Upload
diego-ferioli
View
180
Download
11
Embed Size (px)
DESCRIPTION
M.A. Thesis at Háskóli Íslands - University of Iceland
Citation preview
Hugviacutesindasvieth
The Root Vocalism of the Preterite of two Subclasses of Old
Icelandic VII Class Strong Verbs
Ritgereth til MA-proacutefs
Diego Ferioli
Maiacute 2010
Haacuteskoacuteli Iacuteslands
Hugviacutesindasvieth
Medieval Icelandic Studies
The Root Vocalism of the Preterite of two Subclasses of Old
Icelandic VII Class Strong Verbs
Ritgereth til MA-proacutefs
Diego Ferioli
Kt 1510853269
Leiethbeinandi Haraldur Bernharethsson
Maiacute 2010
Abstract
The goal of this study is to discuss the controversial evolution of class VII
strong preterites in Old Norse which retained traces of the old Proto-Indo-
European and Proto-Germanic reduplicated perfect In particular the focus
will lay on two subgroups of class VII strong verbs in Old Icelandic which
from the 14th century onwards start being written with a diphthong ltiegt in
the preterite root as if from a long vowel (eg hielt fiekk and snieri rieri)
Orthographic evidence from the earliest Old Icelandic manuscripts is then
collected leading to the conclusion that the root vowel in the analysed
preterites forms was clearly a short monophthong (e) in early Old
Icelandic
In light of a review of the theories about the etymology of the preterites of
class VII strong verbs it is then proposed that the root vowel in the
preterites of the mentioned subclasses was short since Proto-Germanic
times and that it arose from the formerly reduplicated syllable after a shift
of the accentuation from the elided root to the reduplicating syllable The
diphthongisation is then traced back to multiple causes A first
phonological diphthongisation took place in words with word-initial h
affecting class VII preterites too (helt hekk) This initial diphthongisation
caused the spreading of the diphthong [je] from other VII class strong
preterites (heacutet greacutet) which had diphthongised because of etymological long
vowel In the modern language preterite plural forms directly derived from
forms with a short vowel are still observable as they show a different kind
of diphthongisation to [ei] (fengum gengum) The preterites of the second
subclass (snera rera etc) adopt the diphthong much later perhaps as late as
the 18th century as a result of their reanalysis as weak verbs and the
neutralisation of the opposition of quantity in the present stem
1
Table of Contents
1 Introduction hellip 4
2 The Old Icelandic Vowel System hellip 7
3 On Reduplication hellip 11
4 Views on the Rise of VII Class Strong Preterites hellip hellip 17
5 On ē2 and the Spreading of [je] in Old Icelandic helliphellip 21
6 The Orthography of the Earliest Manuscripts hellip 27
61 Introduction hellip 27
62 The earliest Icelandic Manuscripts Larsson (1981) hellip 31
63 Holm perg 15 4to hellip 33
64 The earliest Norwegian Manuscripts Holtsmark (1955) hellip 36
65 NRA 52 hellip 38
66 GKS 2087 4to hellip 38
67 AM 519a 4deg hellip 40
68 AM 132 fol hellip 41
69 Summary hellip 43
7 Conclusions hellip 44
8 Bibliography hellip 48
2
List of Abbreviations
Go = Gothic
Icel = Icelandic
IPA = International Phonetic Alphabet
OE = Old English
OFris = Old Frisian
ON = Old Norse
OS = Old Saxon
OHG = Old High German
PIE = Proto-Indo-European
3
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Haacuteskoacuteli Iacuteslands
Hugviacutesindasvieth
Medieval Icelandic Studies
The Root Vocalism of the Preterite of two Subclasses of Old
Icelandic VII Class Strong Verbs
Ritgereth til MA-proacutefs
Diego Ferioli
Kt 1510853269
Leiethbeinandi Haraldur Bernharethsson
Maiacute 2010
Abstract
The goal of this study is to discuss the controversial evolution of class VII
strong preterites in Old Norse which retained traces of the old Proto-Indo-
European and Proto-Germanic reduplicated perfect In particular the focus
will lay on two subgroups of class VII strong verbs in Old Icelandic which
from the 14th century onwards start being written with a diphthong ltiegt in
the preterite root as if from a long vowel (eg hielt fiekk and snieri rieri)
Orthographic evidence from the earliest Old Icelandic manuscripts is then
collected leading to the conclusion that the root vowel in the analysed
preterites forms was clearly a short monophthong (e) in early Old
Icelandic
In light of a review of the theories about the etymology of the preterites of
class VII strong verbs it is then proposed that the root vowel in the
preterites of the mentioned subclasses was short since Proto-Germanic
times and that it arose from the formerly reduplicated syllable after a shift
of the accentuation from the elided root to the reduplicating syllable The
diphthongisation is then traced back to multiple causes A first
phonological diphthongisation took place in words with word-initial h
affecting class VII preterites too (helt hekk) This initial diphthongisation
caused the spreading of the diphthong [je] from other VII class strong
preterites (heacutet greacutet) which had diphthongised because of etymological long
vowel In the modern language preterite plural forms directly derived from
forms with a short vowel are still observable as they show a different kind
of diphthongisation to [ei] (fengum gengum) The preterites of the second
subclass (snera rera etc) adopt the diphthong much later perhaps as late as
the 18th century as a result of their reanalysis as weak verbs and the
neutralisation of the opposition of quantity in the present stem
1
Table of Contents
1 Introduction hellip 4
2 The Old Icelandic Vowel System hellip 7
3 On Reduplication hellip 11
4 Views on the Rise of VII Class Strong Preterites hellip hellip 17
5 On ē2 and the Spreading of [je] in Old Icelandic helliphellip 21
6 The Orthography of the Earliest Manuscripts hellip 27
61 Introduction hellip 27
62 The earliest Icelandic Manuscripts Larsson (1981) hellip 31
63 Holm perg 15 4to hellip 33
64 The earliest Norwegian Manuscripts Holtsmark (1955) hellip 36
65 NRA 52 hellip 38
66 GKS 2087 4to hellip 38
67 AM 519a 4deg hellip 40
68 AM 132 fol hellip 41
69 Summary hellip 43
7 Conclusions hellip 44
8 Bibliography hellip 48
2
List of Abbreviations
Go = Gothic
Icel = Icelandic
IPA = International Phonetic Alphabet
OE = Old English
OFris = Old Frisian
ON = Old Norse
OS = Old Saxon
OHG = Old High German
PIE = Proto-Indo-European
3
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Abstract
The goal of this study is to discuss the controversial evolution of class VII
strong preterites in Old Norse which retained traces of the old Proto-Indo-
European and Proto-Germanic reduplicated perfect In particular the focus
will lay on two subgroups of class VII strong verbs in Old Icelandic which
from the 14th century onwards start being written with a diphthong ltiegt in
the preterite root as if from a long vowel (eg hielt fiekk and snieri rieri)
Orthographic evidence from the earliest Old Icelandic manuscripts is then
collected leading to the conclusion that the root vowel in the analysed
preterites forms was clearly a short monophthong (e) in early Old
Icelandic
In light of a review of the theories about the etymology of the preterites of
class VII strong verbs it is then proposed that the root vowel in the
preterites of the mentioned subclasses was short since Proto-Germanic
times and that it arose from the formerly reduplicated syllable after a shift
of the accentuation from the elided root to the reduplicating syllable The
diphthongisation is then traced back to multiple causes A first
phonological diphthongisation took place in words with word-initial h
affecting class VII preterites too (helt hekk) This initial diphthongisation
caused the spreading of the diphthong [je] from other VII class strong
preterites (heacutet greacutet) which had diphthongised because of etymological long
vowel In the modern language preterite plural forms directly derived from
forms with a short vowel are still observable as they show a different kind
of diphthongisation to [ei] (fengum gengum) The preterites of the second
subclass (snera rera etc) adopt the diphthong much later perhaps as late as
the 18th century as a result of their reanalysis as weak verbs and the
neutralisation of the opposition of quantity in the present stem
1
Table of Contents
1 Introduction hellip 4
2 The Old Icelandic Vowel System hellip 7
3 On Reduplication hellip 11
4 Views on the Rise of VII Class Strong Preterites hellip hellip 17
5 On ē2 and the Spreading of [je] in Old Icelandic helliphellip 21
6 The Orthography of the Earliest Manuscripts hellip 27
61 Introduction hellip 27
62 The earliest Icelandic Manuscripts Larsson (1981) hellip 31
63 Holm perg 15 4to hellip 33
64 The earliest Norwegian Manuscripts Holtsmark (1955) hellip 36
65 NRA 52 hellip 38
66 GKS 2087 4to hellip 38
67 AM 519a 4deg hellip 40
68 AM 132 fol hellip 41
69 Summary hellip 43
7 Conclusions hellip 44
8 Bibliography hellip 48
2
List of Abbreviations
Go = Gothic
Icel = Icelandic
IPA = International Phonetic Alphabet
OE = Old English
OFris = Old Frisian
ON = Old Norse
OS = Old Saxon
OHG = Old High German
PIE = Proto-Indo-European
3
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Table of Contents
1 Introduction hellip 4
2 The Old Icelandic Vowel System hellip 7
3 On Reduplication hellip 11
4 Views on the Rise of VII Class Strong Preterites hellip hellip 17
5 On ē2 and the Spreading of [je] in Old Icelandic helliphellip 21
6 The Orthography of the Earliest Manuscripts hellip 27
61 Introduction hellip 27
62 The earliest Icelandic Manuscripts Larsson (1981) hellip 31
63 Holm perg 15 4to hellip 33
64 The earliest Norwegian Manuscripts Holtsmark (1955) hellip 36
65 NRA 52 hellip 38
66 GKS 2087 4to hellip 38
67 AM 519a 4deg hellip 40
68 AM 132 fol hellip 41
69 Summary hellip 43
7 Conclusions hellip 44
8 Bibliography hellip 48
2
List of Abbreviations
Go = Gothic
Icel = Icelandic
IPA = International Phonetic Alphabet
OE = Old English
OFris = Old Frisian
ON = Old Norse
OS = Old Saxon
OHG = Old High German
PIE = Proto-Indo-European
3
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
List of Abbreviations
Go = Gothic
Icel = Icelandic
IPA = International Phonetic Alphabet
OE = Old English
OFris = Old Frisian
ON = Old Norse
OS = Old Saxon
OHG = Old High German
PIE = Proto-Indo-European
3
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Runar heita geltir en ruacutenar maacutelstafir
First Grammatical Treatise 8616
1 Introduction
The so-called seventh class of strong verbs is a well known puzzle within
Germanic linguistic research The preterite of these verbs evolved from
Proto-Germanic where all strong verbs were once reduplicated1 to the
attested daughter languages where with the notable exception of Gothic
(the only instance of reduplication being a synchronically analysable
phenomenon) very few traces of reduplication are recorded and different
preterite patterns are found instead Some of them present a vowel
alternation which could be synchronically explained as a divergent kind of
ablaut some others feature what could either be infixation or the result of a
contraction of the root and the ancient reduplicating syllable Such verbs are
attested in all early Germanic languages the best attested ones being
Gothic Old Norse Old English Old High German Old Saxon and Old
Frisian although it is evident that at the time of the attestation
reduplication was being abandoned for good as a regular way of forming
the preterite
Old Norse retains a handful of interesting instances of formerly
reduplicated preterites The verbs examined here constitute an important
part of the core common Germanic vocabulary preserved in Old Norse and
show an interesting mixture of archaic traits and innovations The main
difficulty for the linguist attempting to trace the development of these
1 See Bammesberger 199415ff Jasanoff 2003168 ff
4
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
forms is their great variety of patterns within a fairly small verb class The
following six structurally defined subclasses have traditionally been
presented for classical early 13th-century Icelandic (see for instance Noreen
1970337-339 for the subdivision subclass VIIe only entry bloacuteta is most of
the times grouped with VIId the glosses are reported as according to
Cleasby-Vigfuacutesson 1874)
Table 1 The Old Icelandic VII Class of Strong Verbs
Subclass Vowel Alternation
Infinitive English Translation
3p pret sg 3p pret pl
VII a ei - ē heita be called command
heacutet heacutetu
leika play leacutek leacuteku
VII b au - jō hlaupa run hljoacutep hlupu
auka add joacutek
ausa pour joacutes josu
houmlggva chop hjoacute hjoggu hjuggu
ū - jō buacutea dwell bjoacute bjoggu bjoumlggu bjuggu
VII c a - e halda hold helt heldu
falla fall fell fellu
falda fold felt feldu
blanda blend blett blendu
ganga walk gekk gengu
hanga hang hekk hengu
faacute (lt fanhaną) get fekk fengu
VII d ā - ē raacuteetha rule reacuteeth reacuteethu
blaacutesa blow bleacutes bleacutesu
graacuteta weep greacutet greacutetu
laacuteta let leacutet leacutetu
VII e ō - ē bloacuteta bleed bleacutet bleacutetu
VII f ū - e snuacutea turn snoslashra snera sneru
gnuacutea rub gnoslashra gnera gnoslashru gneru
ō - e roacutea row roslashra rera roslashru reru
groacutea grow groslashra grera groslashru greru
ā - e saacute sow soslashra sera soslashru seru
5
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Moreover Noreen (1923338-339) lists sveipa (pret sveip) in subclass VIIa
and mentions two isolated participles eikinn and aldinn which could fit
into subclasses VIIa and VIIc respectively The verb bnuacutea in VIIf (attested
only in the preterite bnera) may be a variant of gnuacutea
From the subdivision above it is clear that all of these preterite singular
forms except subclass VIIf (sneri reri etc) are monosyllabic and except
VIIc (helt fekk etc) and again VIIf all have a long root vowel This study
will especially focus on these two latter sub-classes as in the course of the
14th century for subclass VIIc and after the 16th century for subclass VIIf
the root vowel in the preterite usually believed to once be have been short
surfaces as a diphthong written ltiegt just as if it had come from a long
vowel (ē) The effects of such sound change are still very well observable
in modern Icelandic where the spelling lteacutegt reflects a [je][je]
pronunciation The dynamics according to which this unusual
diphthongisation took place are still unknown and such lack of knowledge
constitutes a major obstacle towards understanding the actual development
of this class of verbs into modern Icelandic
As it will be shown vowel length is not indicated systematically in Old
Icelandic written sources That the vowel system was internally distinctively
divided by the feature of length seems to have been well known even to
early 12th-century Icelanders as clearly confirmed by the First Grammatical
Treatise (see below) Individual scribes however may oftentimes fail to
mark vowel length by avoiding it completely or applying it inconsistently
Moreover contamination among divergent scribal practices and
interpolation of manuscripts during various stages of the tradition may
obscure the picture presented by the orthography even further Before
tackling the problem of the orthography a few considerations about the
6
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
historical development of the Old Icelandic vowel system and reduplication
are to be made
2 The Old Icelandic Vowel System
Old Icelandic retained an ancient Proto-Indo-European vowel length
pattern ie syllable length could be of three types light (VC) heavy
(VC or VC) and hyper-characterised (VC) Vowel length was
originally contrastive and independent from any other variable The Old
Icelandic vowel system was derived from Proto-Germanic after being
heavily restructured by syncope at a Proto-Norse stage which re-shaped
vowel length in unstressed syllables and triggered the phonemicisation of
the new vowels arising from umlaut processes ultimately greatly enlarging
the inherited phoneme inventory (cf Garnes 1976196-199 )
Table 2 - The Proto-Norse Vowel System
SHORT LONG DIPHTHONGS
Front Back Front Back
High i u ī ū eu [ju]
Mid e ē2 ai
Low a ē1 ō au
7
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Table 3 The Common Norse Vowel System after Syncope
SHORT LONG NASAL2 DIPHTHONGS
front back front back front back
High i y u iacute yacute uacute ĩ y ũ ey
Mid e oslash o eacute ǿ oacute ẽ oslash otilde ei
Low aelig a ǫ ǽ aacute ǫ aelig atilde ǫ au
Table 4 The classical Old Icelandic Vowel System (ca 1200)
SHORT LONG DIPHTHONGS
front back front back
High i y u iacute yacute uacute ey
Mid e oslash o eacute ǿ oacute ei
Low a ǫ ǽ aacute au
Around the time of the First Grammatical Treatise (mid-12th century) the
vowel system was still divided into three main subsystems of short long
and nasal vowels (table 3) Possibly because the short and long subsystems
were not symmetrical a significant re-shuffling gradually took place in
several steps over a long period of time in the history of Old Icelandic and
eventually the distinctive feature of quantity was eliminated to the benefit
of quality Firstly the phonemic distinction between oslash and ǫ which
merged into ouml (ca 1200 or shortly thereafter) was neutralised secondly
the merger of aelig and ǿ into aelig took place (ca 1250) later long mid and
low vowels became diphthongs and syllable types were reduced from four
2 According to Hreinn Benediktsson (1972128-137) there is enough evidence pointing at fully distinctive nasal correlation at the time of the First Grammatical Treatise Being nasality and length superimposable features a more accurate representation would have showed a distinction between long and short nasal vowels however due to the complementary distribution which seems to apply to nasal and non-nasal short vowels and the fact that the nasality correlation was neutralised when long vowels when a nasal consonant followed I have opted for a simpler chart showing the nasal correlation as a third type of vowel correlation besides length and shortness
8
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
to two (ie VC and VC) so that all stressed syllables became heavy
Whereas in the mid and high vowel subsystems there had nearly always
been a one-to-one correspondence between long and short the low vowels
rather exhibited an opposition of frontness (note the shift of aacute from a
central to a back position after the disappearance of ǫ)3 The abandonment
of contrastive vowel length is commonly referred to as the Icelandic
Quantity Shift (Icel hljoacuteethvalarbreytingin) stretching at least from the 13th to
the 16th century (cf Bjoumlrn K THORNoacuteroacutelfsson 1929a Aacuternason 1980121ff and
2005332ff) Such phenomenon reflects a wider tendency to re-organise a
languages vowel inventory in terms of quantity and emergence of new
diphthongs which is common to most other old Germanic languages (see
for instance Haugen 198224ff on Scandinavian and Prokosch 1939107)
The diphthongisation in the late 13th and 14th centuries brought about a
new in a way less definite dicotomy between unimoric and bimoric vowels
rather than between long and short vowels by means of a glide added to the
former long vowels coming to constitute the reshaped second mora To the
front vowels the front semi-vowel i was added (aelig [aelig] gt [aeligi] gt [ai])
while a back semi-vowel u was added to the back vowels (aacute [a] gt [au])
As early as in the 13th century however the new realisation of eacute [ei]4 came
to be confused with the older diphthong ei (eg in minimal pairs such as
meir and meacuter) so that the process was soon reversed to [ie] (cf Hreinn
3 The issue is in fact still open to question and it is not to be excluded that the merger of aacute and ǫ may have resulted in a round vowel not affecting its backnessfrontness
4 Others explain ei as dialectical as it is apparently not found in Western Iceland (Noreen 192395) However considering that all new diphthongs are descending with [je] being the only exception the descending [ei] could have well been its predecessor Unclear is the development of aelig into [ai] if the early spelling ltiaeliggt reflects an ascending diphthong the phoneme might have undergone the opposite change ie from ascending to descending But again that might be a regional variant from Northern Iceland and in part Breiethafjoumlrethur (Aacuternason 2005333)
9
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Benediktsson 1959298 and Kristjaacuten Aacuternason 2005333) It follows that the
glide must have been originally added to the high vowels as well to which it
was identical in quantity the result was very close to the original value of
the former long vowels [i] ~ [ii] [u] ~ [uu] Although it has never been
proposed before in fact it is not to be excluded that the glide spread from
the high vowels down to the low vowels implying a reanalysis of the
constituents of the two morae (cf Table 5) Down to the modern language
former mid and low long vowels are best analysed as a sequence of two
separate phonemes (especially [je] as j+e) while the high vowels
(including y and yacute which eventually merged with i from a certain
point on lost any bimoric manifestation
Table 5 The Late Old Icelandic Vowel System (ca 1200-1500)
UNIMORIC BIMORIC DIPHTHONGS PROPER
front back front back
High i y u iacute [ii] uacute [uu] ey
Mid e ouml o eacute [ei]gt[ie] oacute [ou] ei
Low a aelig [ai] aacute [au] au
Such transformations are indeed very complex and as one can expect took a
considerable amount of time to become established both in the spoken and
by reflex in the written language The only way to determine how the
changes took place is thorough a orthographic investigation not forgetting
that the manuscripts preserved are mostly neither consistent nor do they
always represent a single phase of the history of the language they are
written in Nevertheless as will be confirmed later in greater details the
data just examined above show a clear tendency towards the replacement of
length as a distinctive feature by quality possibly as early as the 13th
century and lasting until the 16th century Garnes (1976198) defines the
10
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Quantity Shift as ldquoan increase of the scope of the suprasegmental feature
quantityrdquo meaning that before the shift ldquothe scope of the quantity was the
segment whereas in the post-quantity shift period the scope was the
syllablerdquo
3 On Reduplication
Reduplication originates in Proto-Indo-European as a morphological tool
to form the perfect The perfect denoted a stative aspect meaning that it
expressed a state resulting from the completion of an action and was
characterised by e-reduplication plus mainly o zero apophony plus
special endings partly recalling those of the middle and hysterokinetic
accentuation (Jasanoff 200330 and 2007242) Later the Proto-Indo-
European (active) perfect became a simple past tense in Germanic and was
further specialised to the point that all strong Germanic preterites must
once have been reduplicated Later on however except for subclass VIIf
there are no clear vestiges of reduplication in Old Norse and most of its
relatives making the loss of this peculiar and once very productive
morphological feature one the most important innovations in the early
history of the Germanic languages
The next steps were the rise of a formal dichotomy between reduplication
and ablaut formerly superimposable features as two concurring
alternatives for forming the preterite The reduplicating syllable was not
stressed in the Proto-Indo-European perfect and it is generally assumed
that it never was in Proto-Germanic either (Ringe 2007191) Because of the
accent lying on the root vowel one might expect Verners Law to have
voiced root-initial fricatives in those stems that were still reduplicating sure
instances of that are very few (eg ON sera (s)he sowed Go gasaiacutezlep
11
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
(s)he had fallen asleep) but of a particular relevance since their reanalysis
gave rise to a -Vr- infix which according to some was extended to a good
number of other stems in Old Norse (where sera alone gave birth to
subclass VIIf) and Old High German Since the voicing s gt z is the only
example of Verners Law in class VII verbs it may be concluded that other
alternations of root-initial fricatives had already been levelled in Proto-
Germanic (Ringe 2007191-192)
The main difference between reduplication in Gothic and Northwest
Germanic is the treatment of the initial clusters Obstruent + sonorant
clusters reduplicate in Gothic with the obstruent only (cf slepan ~ saiacuteslep
fraisan ~ faiacutefrais) while NWGmc retains the cluster initially but simplifies
it medially (cf ON grera OE -dreord OHG pleruzzun) However Gothic
does not entirely reflect the Proto-Germanic situation Gothic innovates
from Proto-Germanic in the following aspects a) neutralisation of Verners
Laws effects (with few exceptions) b) weak plural (and dual) stem forms
are levelled in favour of the singular and c) generalisation of -ai- [ɛ] at the
expense of -i- (cf Jasanoff 2007244)
While most preterite roots are believed to have somehow been re-shaped by
contraction with the reduplicating syllable there is a subgroup which may
simply have dropped it The reason for this development was that the roots
were already ablauting so that the reduplicating syllable became redundant
as a mark for the past tense There are at least eight such verbs which
retained both reduplication and ablaut possibly until a late stage in Proto-
Germanic
blēsaną blow ~ beblōsgrētaną weep ~ gegrōt hwētaną push continuously ~ hehwōt
12
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
lētaną let ~ lelōt letrēdaną rule~ rerōd ruletēkaną take ~ tetōk takesēaną sow ~ sezōwēaną blow (of wind) ~ wewō
Ringe (2007250) hints at more strong verbs with root-internal ē which
are attested elsewhere in the Germanic family tree Some of those did
probably not ablaut like slēpaną ~ sezlēp Only two of these verbs surely
kept the original o-vocalism in the preterite unaltered In Old Norse a
considerable part of them migrated into other classes or were relocated
within class VII wēaną fell out of use sēaną remained basically
unchanged but was relocated with the non-ablauting rōaną to form
subclass VIIf tēkaną and hwētaną which meanwhile both developed a
short instead of a long a-vocalism in the infinitive as taka became the
former a simple class VI strong verb as to hwētaną it apparently surfaces
as a weak verb5 Last but not least the remaining preterites changed their
root vocalism from ō to ē but not without leaving traces in Old
Swedish loacutet and of course sezō which re-emerged after the syncope as a
reanalysed sera with regressed accentuation (Ringe 2007249)
The developments which led to the new VII class of strong verbs can be
now summarised as follows
a) rise of a new generalised ē-vocalism (of unclear origins) marking the
preterite tense and applying to subclasses VIIa VIId and VIIe
5 The attestation of hwētaną is problematic It is found both as the weak verbs hvata and hvota in Old Icelandic where hvota (seemingly from hvaacuteta) seems to regularly derive from hwētaną and is probably related to the other weak verb hoacuteta (to hold forth with threatening gestures) which probably merged with older hǿta (to threaten) the form hvata seems to suggest a development similar to taka (cf Ringe 2007249 CleasbyVigfuacutesson 1957297 and 281)
13
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
b) tendency toward monosyllabic preterites matching the present forms
and consequently
c) elimination of either the reduplicating syllable or the root syllable in
the above subclasses (see chapter 4)
d) in subclass VIIb especially when the root started with a vowel a kind
of contraction took place at the same time re-modelling the new ō-
vocalism to a pattern close if not identical to class VI preterites (the
migration of taka to class VI confirms the analogy)
e) subclass VIIf would then work as a waste bin for orphaned forms
which because of their inner structure fully retained reduplication
but of which only sera did originally ablaut the last vowel was then
easily reanalysed as a weak ending -a analogous to -etha
f) the verbum puro būaną did not join subclass VIIf in most Germanic
languages (unlike Old High German birun) and was relocated to
subclass VIIb after undergoing contraction būaną ~ beƀ gt
bew gt beū gt bjoacute (plural forms would then be analogical
according to the VIIb alternation)
The question that needs to be answered before dealing with the
reconstruction of class VII strong preterites is whether the root vowel was
long or short This is the case for especially subclass VIIc preterites since
even when their development is satisfactorily explained the results are
often hardly applicable to the remaining subclasses In Proto-Germanic
subclass VIIc verbs used to form the preterite in roughly the same way as
subclass VIIa verbs so they are expected to behave in a similar way even
after the reduplication period And yet their development is different as
they do not seem to be drawn (yet) towards the newly created ē-grade It
has been maintained that the reason for this is to be sought in their inner
structure (see chapter 4) Their distinctive features are the following
14
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
a) their default (present) root structure is CaRC rather than
CeiC
b) the final consonant was originally voiced and if it is not a
resonant it is later devoiced and if there is an adjacent nasal it is
assimilated
c) at times not only the quantity but also the quality of the root
vowel fluctuates between e and i in Old Norse (i in East Norse
e in West Norse although i is sometimes found as a variant in
Old Icelandic cf Chapter 6 section on Moumlethruvallaboacutek) Old Saxon
and Old Frisian
An overview of the preterite singular forms is given below in table 6 (cf
Fulk 1987169-172 Torp 1909 Katara 1939 for Old Saxon Steller 1928 and
Bremmer 2009 for Old Frisian)
Table 6 Comparative Overview of Subclass VIIc Preterite Singular Forms
ON OHG OS OE OFris Go
held hialt held hēold hēldhīld haiacutehald
felt fialt feld fēold faiacutefald
fekk fiang feng fēng fengfing faiacutefāh
hekk hiang heng hēng henghweng haiacutehāh
gekk giang geng gēong gengging (gaiacutegagg)
fell fial fellfēl fēoll fol faiacutefal
Old High German shows a clear long vowel since its earliest period (ē
later ia and then ie) alongside Old English where it is more
problematic to trace back the original vowel length but it seems however
that the given diphthongs probably came from long vowels (Fulk 1987171)
As shown in the table in both Old English and Old High German the VIIc-
15
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
type has been absorbed into another type characterised by a long vowel in
the preterite In Old English it merged with the b-type which was
productive and serving as a box for verbs with divergent vocalisms in Old
High German it merged with the a-type acquiring thus the diphthong ia
in the preterite Through this process the anomaly of a short vowel in class
VII was soon obliterated in West Germanic (Joumlrundur Hilmarson 199138-
39) with the notable exceptions of the manuscripts of the so-called Isidore
Group and the Monsee-Wiener fragments (Braune 1967286-287) and Old
Saxon The latter shows chiefly a short vowel being thus all in all identical
to Old Icelandic The vowel can occasionally be lengthened or
diphthongised before a simple consonant like in fell gt fel gt fēl but it
definitely appears to be a later minor innovation The evidence for Old
Frisian is somewhat less clear due to an extensive reshuffling of the vowel
system although a short vowel seems to dominate
Postulating an early long vowel in all Northwest Germanic remains an
option but has further implications Old Icelandic is not the anomaly the
diphthongisation e gt ei gt ie which seems to affect some of subclass
VIIc preterites is found not only in Old Icelandic but remarkably also in
Old Saxon and Old Frisian sources in apparently the same instances
Particularly in the oldest Old Saxon attestations (the manuscripts of the
Heliand) the root vowel in class VIIc is nearly always noted as short and
fluctuates between e and i (as in gengging) whereas the latter phoneme
is stabilised in East Norse at about the same time In late Old Saxon it also
starts appearing as lteegt lteigt lteygt and in those words containing i as
ltiegt and later into Middle Low German it monophthongises to lteegt (cf
Katara 1939114) There is no apparent reason for the fluctuation e~i to
be dependent on a long vowel or even less plausibly to be a reflex of
Proto-Germanic ē2 (as it is often claimed among others by Fulk
16
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
1987171) and why Old English and Old High Germans should not have
been the innovators In fact the situation looks rather like the preservation
of an archaism which has later been analogically levelled There is in short
no need of postulating a long chain of lengthenings and shortenings in
order to explain why Old Norse was different from the rest rather the other
way around In the following chapter theories on the formation of class VII
preterites will be examined more specifically
4 Views on the rise of VII class strong preterites
Traditional handbooks of Old Norse grammar display different and at
times inconsistent views on the development of reduplicating verbs In one
of his early works Adolf Noreen (1913205) delineates a twofold pattern
following a tradition which had started already with Grimm
a) fefall gt ffall gt ON fal(l) from inf falla
b) hehald gt hēalt gt ON heacutelt from inf halda
Noreen believed that two different changes occurred within subclass VIIc
and that pattern b) was later generalised to a) which originally retained the
root vowel quality due to the intervening fricative As a consequence of such
analogical change most subclass VIIc preterites would then have had a long
vowel from the beginning resulting partly from a compensatory
lengthening (hehald gt hēalt) which later causes a contraction of the root
vowel with the one in reduplicating syllable and partly on analogy The
diphthongisation in Icelandic would then be due to the long monophthongs
whereas in subclass VIIf the root vowel was lengthened because in an open
17
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
syllable (sneri gt sneacuteri) Those forms which were not subject to this change
underwent analogical change (fall gt feacutell)
Heusler (195092-93) produced a much more modern-looking picture with
a short vowel in all subclass VIIc preterites He argued that ldquodas nord fell
verlangt keine Vorstufe fēllrdquo His intention was possibly to criticise the
common idea that a long vowel had arise from contraction or secondary
lengthening and then later been shortened as in Boer (1920191) ldquoDe e is
door verkorting uit ē (ē₂) ontstaanrdquo
An outdated approach that for a time met with widespread approval was a
phono-morphological ablaut-based derivation once popular among the
Neogrammarians This has later turned out not to be a feasible approach
especially after the advent of the laryngeal theory An epitome of this school
is Van Coetsem (1956) Van Coetsem pointed out that pure reduplication as
in haiacutehait is exclusively found in Gothic while Northwest Germanic
developed its own VII-class ablaut grades In his scenario subclasses VIIa
(present root vocalism ai) and VIId (present root vowel ē1) had
originally developed the same ldquoaugmented graderdquo ei in the preterite
later this newly formed diphthong ei monophthongised to ē2 pushed
by the development ai gt ei in the infinitive Subclass VIIc fits nicely
in to the reconstructed alternation ai ~ ei and similarly in VIIb au
~ eu (gt jō) corresponds the simpler a ~ e This kind of
alternation has also been called reversed ablaut (Ablaut in umgekehrte
Richtung) due to the mirrored image of the Proto-Indo-European e ~
a ablaut alternation This theory still has much to recommend it
especially because by acknowledging an original short e in subclass VIIc
preterites the overall picture becomes simpler
18
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
However the approach which eventually has had most success is strictly
phonological In short it is a theory of contraction of the formerly
reduplicating syllable and the root syllable similar to the one formulated by
Noreen and discussed above Contraction generates a new kind of root
vocalism that is ultimately reanalysed as a new ablaut type Scholars do not
agree on the exact details of the contraction and presently divide between
those advocating a mixture of syncope and compensatory lengthening
infixation ldquocompressionrdquo and haplology
Voyles (199273) explains how ldquoone of the ways class VII strong verbs
formed the past was to insert stressed e into the present-tense stemrdquo
Although subclass VIIc is not specifically discussed one can assume that
Voyles believes Old Norse helt to have been hēlt since the beginning and
having developed from hald gt heald gt healt gt hēlt Voyles does not
produce any real sound law to explain the change ea gt ē only that ldquothe
new diphthong eē in forms like leēt let was interpreted quite naturally
as ērdquo and so ldquosimilarly the new diphthong ea [hellip] was also interpreted as
ērdquo Even though most of this is actually in line with earlier theories such
developments are in fact highly implausible and not supported by any
actual orthographic evidence (ie the Old Icelandic written sources indicate
that subclass VIIc verbs had a short root vowel in the preterite as will be
discussed below) It is also likely that some stage of this process assumed by
Voyles and the Old Norse breaking were overlapping at some point in
North Germanic and given the similarity between a hypothetical
diphthong ea in the preterite stem and the first stage of breaking (e
gt ea gt ja) one would rather expect the development ea gt ja as in
the breaking environment Arguably this was most probably the context
from which the preterite joacutek arose from eauk
19
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Another kind of infixation theory has been proposed by Vennemann
(1994) According to Vennemann the different root syllables in
reduplicated forms typically made up of a fricative and a vowel were first
generalised to -zV- and later rhotacised to -rV- Later the root vowel was
syncopated and stress was stabilised on the reduplicating syllable somehow
opening the way for the elision of -r- which triggered a compensatory
lengthening of the stress-bearing vowel Vennemann assumes therefore that
subclass VIIc preterites got a long vowel at an early stage The assumption
that there ever was a stress regression from the root to the reduplicating
syllable appears to be peculiar of Vennemannrsquos theory (1994306-307)
Vennemannrsquos theory has been firmly criticised by Jasanoff (2007) who also
notably claims that most of Vennemannrsquos sound laws have been produced
ad hoc Jasanoff maintains that the word-initial (ie radical) stress was
established on a Proto-Germanic level thus including Gothic as well and
that long vowels were not shortened in unstressed position in either North
and West Germanic concluding therefore that ldquothere is no way in short
that the position of the accent could have been a determining factor in the
restructuring of reduplication in Northwest Germanicrdquo (Jasanoff 2007257)
Instead he reformulates the contraction theory into a new-looking
ldquocompressionrdquo theory where disyllabic preterite plural forms were made
monosyllabic through the ldquoejectionrdquo of the root vowel first in the preterite
plural the consequent cluster was then simplified and the new vowel
alternations were reanalysed as ablaut As a necessary criticism it should be
noted that Jasanoff does not clarifies how the ldquoejectionrdquo could have taken
place without the regression of the stress from the root to the reduplicating
syllable however
20
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
5 On ē2 and the spreading of [je] in Old Icelandic
Class VII preterites such as heacutet and reacuteeth are usually reconstructed with
Proto-Germanic ē2 a late phoneme resulting from the lengthening of
short e and distinct from Proto-Germanic primary ē (or ēsup1 lt PIE
ē) There can be little doubt that in order to be two different phonemes
ēsup1 and ē2 must have had some more or less marked difference in
quality Evidence for ēsup1 being rather a low-spread vowel perhaps better
noted with ǣ have been convincingly produced by Antonsen (200220)
that is evident from its later development to long ā in stressed syllables from
the lowering influence it has on preceding root vowels as when
P[roto]G[ermanic] klibǣnan gt OHG klebēn Ger kleben to cleave to stick
and from early foreign loanwords in which a foreign long mid-front vowel ē is
borrowed into Germanic as the long high front vowel ī [hellip] This sound-
substitution in loanwords occurred because there was no corresponding long mid
vowel in Germanic
According to Joumlrundur Hilmarson (1991) Proto-Germanic ē2 as a
distinctive phoneme is primarily found in the word for here which he
derives from the Proto-Indo-European particle ke (rather than from the
instrumental singular kī) whose e was ldquoemphaticallyrdquo lengthened and
to which an analogical -r was added In addition Joumlrundur Hilmarson
assumes that ē2 was found in most preterites of class VII strong verbs It
is also found in some other minor instances most notably in Latin
loanwords containing ē The phoneme is therefore a Germanic
innovation and surfaces whenever lengthening occurs
The attestation of Old Icelandic heacuter and that of VII class strong preterites
does indeed have some important similarities which necessitated the
21
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
inclusion of the adverb in this present study Joumlrundur Hilmarson did his
own survey of the orthographic manifestation of herheacuter on the Old
Icelandic Book of Homilies Holm Perg 15 4to (about 1200) He states
(199134)
According to Hreinn Benediktsson (19676832 [in the bibliography of this
present study listed as Hreinn Benediktsson 1967]) over 90 of long vowels in
some of the hands of that manuscript exhibit the length mark The adverb ldquohererdquo
however when written in full has a short vowel in all of its 44 occurrences and
when abbreviated it is written with a length mark only once out of 39
occurrences This seems to necessitate the assumption that Old Icelandic
possessed not only the form heacuter as usually posited and preserved in Modern
Icelandic but also the form her with a short vowel [hellip]
This theory clearly casts new light upon the problem of lengthening of e
in Old Icelandic The attested length of the word here fluctuates for a
time in a fashion that is indeed very similar to some of the preterites of the
VII class of strong verbs and its occurrence is directly observable in those
sources which are also useful for observing the formerly reduplicating
preterites Apparently there are several instances of ē2 over a longer time
span and they are due to multiple sound changes about which very little is
known According to Joumlrundur Hilmarsson (199139-40) the sound
changes giving rise to ē2 apply to three cases
bull Strong verbs of subclass VIIa he-hait- gt heacute-hāt- gt het- (haplologic
elision of the root syllable) gt hē2t- (secondary lengthening analogy
wit the present stem)
bull Similarly in strong verbs of subclass VIId le-lē1t- gt leacute-lāt- gt let- gt
lē2t-
bull her gt hē2r (emphatic lengthening)
22
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Joumlrundur Hilmarsons theory is not far from Jasanoffs compression theory
But while Jasanoff still heavily relies on contraction consequent to the
elision of the root vowel the first employs haplology in order to eliminate
the whole root syllable While on one hand this could very well apply to
Old Norse on the other hand limited evidence of cognate vestiges such as
Anglian heht reord and leort constitute a difficult problem as they seem to
retain clear traces of the root syllable which could therefore not have been
elided at once (see for example Jasanoff 2007245) Moreover a reasonable
amount of caution is needed when it comes to phenomena such as emphatic
lengthening and most importantly haplology in the latter case elision by
haplology calls for the assumption that reduplication as a morphological
tool was entirely no longer understood by the users of the language at the
time The reason for that would not necessarily be that the mechanism had
been made obscure by independent sound changes although it was clearly
no longer productive morphological reduplication must have been liable of
being fully analysed as such but for some reason it simply was not
The exact evolution of reduplicating verbs aside however there seems to be
an evident limitation with most of the linguistic research discussed above in
that its focus generally lies on the quantity of the root vowel in the
discussed preterites and rarely on the diphthongisation which is the actual
recorded phenomenon in several Germanic languages In Old Icelandic in
particular there is enough reason to pay deeper attention at the emergence
of the diphthong [je] in class VII strong verbs but also in a number of other
cases These other cases are so relevant that from the time of the Quantity
Shift on they may even justify analogical transformations within class VII
strong verbs
The first one which needs to be mentioned here involves e gt [je] in words
such as eacuteg eacuteta eacutel heacuteri heacuteraeth heacuteethan Heacuteethinn occurring long before the
23
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
change had started affecting class VII strong preterites (according to Bjoumlrn
Karel THORNoacuteroacutelfsson 1929b223 manuscript sources start reading ltiegt in such
instances from around 1300) The sound change behind these instances has
been explained in several different ways and very often as lengthening with
subsequent diphthongisation whereas both the etymological and
orthographic evidence points at a diphthongisation of a short vowel
Another popular explanation is that [je] arose by means of a palatalisation or
j-infix (cf discussion by Haukur THORNorgeirsson 2009) which in itself is not
very explanatory Theories of the origin of [je] in eacuteg eacuteta etc often include
class VII strong preterites while it is evident that we are dealing with not
only two but more separate phenomena which however produce the same
result and which possibly triggered analogical transformations In words
such as heacuteraeth heacuteethan and the personal name Heacuteethinn for example the change
e gt [je] is triggered by the word-initial h and affects e as from Proto-
Germanic (not resulting from earlier aelig from i-mutated short a)
(Noreen 192395) To class V preterite eacuteta on the other hand possibly a
particular analogical change applies triggered by the preterite singular aacutet
(quite exceptionally both the singular and the plural preterite forms have a
long vowel from a lengthened ablaut grade ē1t-) Finally the adverb heacuter
is most likely one of those words where an initially occasional prosody-
regulated emphatic lengthening could very well apply and later become
generalised These instances albeit not numerous are crucial for the
occurrence of the diphthong [je] as they are part of a varied group of highly
frequent lexemes undergoing the same unusual kind of diphthongisation
Another kind of diphthongisation of short vowels taking place during the
Quantity Shift this time analogical affecting shortened vowels in originally
hyper-characterised syllables This change is limited to the nominal and
adjectival domain and it is another exception to the rule stating that the
24
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
exclusive phonetic condition for the diphthongisation was the long duration
of the vowels Some examples are neuter forms of adjectives such as skjoacutett
fljoacutett braacutett which developed an ldquounderlying diphthongrdquo from an o that
must have been shortened earlier at around 1200 (as in gott ~ goacuteethur and
minn ~ miacuten) because of the evident morphological relationship with their
masculine and feminine equivalents which all had oacute (Kristjaacuten Aacuternason
1980116-118) According to Kristjaacuten Aacuternason one could also consider the
shortened long vowels to have been qualitatively different from the original
short ones so that they could be classified as short allophones of long
vowels (ibid) It should be however noted that such claims are impossible to
substantiate ie there I no way to find out whether there ever was a
lengthening before the diphthongisation because the only valuable sorce
metrics does not distinguish between long and hyper-characterised
syllables Clearly Kristjaacuten Aacuternason excludes that there ever was any
lengthening (1980118)
Furthermore there are also quite a few very interesting examples of
neutralisation of the quantity correlation in Old Icelandic although
affecting very specific instances In monosyllables where a compensatory
lengthening is usually supposed to have taken place the vowel is sometimes
apparently noted as short before junctures (notably in the imperative se thorno
vs seacutethorno (you) see and the phrase Gothornroslashthorne vs goacutethorn roslashthorne in the First
Grammatical Treatise 863-5) Hreinn Benediktsson (1972138) explains
the incongruence as not being an actual shortening for which there would
be no comparable cases but rather a case of vowel length being neutralised
to a point in which the vowel segments in this case taking on the value of
archiphonemes were perceived as neither long nor short This
phenomenon clearly caused fluctuation in pronunciation with the
possibility that the ldquoactual length was more similar to that of the
25
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
phonemically short vowels ie of the negative term of the quantity
correlationrdquo (Hreinn Benediktsson 1972139) The implications of such
observation are manifold but one in particular is interesting for this present
study that
a phonemic change has to be interpolated between Old and Modern Icelandic in
order to account for the fact that the archiphonemes in spite of being identified
with the short phonemes in the twelfth century have later developed in the same
way as the long vowels in positions of distinction (Hreinn Benediktsson
1972139)
which is to say they developed a diphthong (se gt [sje]) directly from a
situation of seemingly short vocalism without previously undergoing the
required lengthening This could definitely account for a further example of
vowel length fluctuation affecting important areas of speech and
ultimately of a diphthong sprung from a vowel which if not surely short
was somehow perceived to be short and certainly noted as such in certain
manuscript tradition It profiles itself as a powerful trend initiated by a
large array of different causes and stretching throughout a considerably long
period of time and gradually extending to a fairly large portion of the
language well into the modern language
Hreinn Benediktssons neutralisation theory becomes very useful when
trying to make sense of the diphthong in the present stem of class V strong
verb eacuteta Criticism is strong on this point Sturtevant (1953457) points out
that if eacuteta is an analogical form it would be the only example of such
analogy in the verbal system whereas on the other hand verbs such as vega
~ vaacute fregna ~ fraacute remain unchanged There are at two main reasons why I
prefer analogy in this particular case One is that it is harder to demonstrate
how often a verb such as eacuteta would occur in contexts where its vowel was
26
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
liable to be lengthened due to its initial position than it is for words such as
eacuteg and heacuter ie leaving the emphatic lengthening hypothesis and the
obviously non-related case of diphthongisation following word-initial h
little explanatory material is left but plain analogy The other one answers
the question as to what kind of analogy this may be taking into account
that preterites of the type of vaacute and fraacute are monosyllables ending with their
stem vowel and were therefore subject to neutralisation of the quantity
correlation it becomes evident that that is the reason why an analogical
infinitive such as veacutega is impossible whereas the diphthongisation of eacuteta
from non-vowel-final aacutet is Also an analogy of the kind gaf aacutet gefa eacuteta
is not entirely out of the question Following this track it should also be
noted that neither the preterites of the verba pura of subclass VIIf (snuacutea
roacutea groacutea nuacutea saacute but also the displaced buacutea) show an unambiguous long
root vowel until late because in early Old Icelandic length in their present
stems was likewise neutralised
6 The Orthography of the Earliest Icelandic Manuscripts
61 Introduction
As discussed above there is some disagreement among scholars concerning
the quantity of the root vowel in subclasses VIIc and VIIf strong preterites
in Old Norse This calls for an inspection of the available sources ie early
Icelandic manuscripts until the 14th century in order to shed some light on
the use of the accent mark and other ways of marking length andor
diphthongisation of eacute Hreinn Benediktsson (196832) and later Kristjaacuten
Aacuternason (1980101) identified only three Old Icelandic manuscripts
containing the most reliable orthographic manifestation of vowel length of
27
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
which only two are of considerable length These three manuscripts are
Holm perg 15 4to NRA 52 and GKS 2087 4to (description below) and
here well represented in this present study but admittedly they did not
contain many relevant examples as will become more evident from a closer
glance
The survey will mainly focus on manuscripts containing prose texts I the
earliest Icelandic manuscripts the principal means of denoting vowel length
is by means of the superscript acute accent mark However the acute mark
has been employed in West Norse for denoting several further grapho-
phonological features besides vowel length and its usage changes
considerably in time and from scribe to scribe Some scholars have even
gone as far as saying that the accent mark is not uniformly used as a length
mark in any manuscript (Seip 195463) Its most common usages are a)
marking a words primary stress especially in words of foreign origin b) as
a diacritic for marking vowels especially on ltigt and ltygt (also noted as
ltijgt) for better distinguishing them from nearby strokes (ultimately
evolving into the dots over i and j) c) as a double mark on two adjacent
vowels it serves as diaeresis (cf Noreen 1913sect33) Other notable practices
are d) marking monosyllables such as aacute iacute oacuter ǫn etc evidently in order not
to confuse them with parts of nearby longer words (however the same
monosyllables are elsewhere often never marked when long unlike long
vowels in other contexts) e) marking vowel quality rather than quantity as
in ltoacutegt for ǫ or oslash and lteacutegt for ę (Seip 195463)
In spite of this it is generally accepted that a primary function of the acute
mark is a quantitative one ie of marking vowel length especially as a
scribal practice which gets established over time According to Lindblad
(1952133-138) it is in the so-called classical period (ca 1150-1350) that the
28
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
length-marking practice is best exemplified reaching peaks of 90 (GKS
2087 4to) while starting around 1250 and throughout the 14th century this
practice yields gradually to the graphical function with a transitional period
stretching roughly from 1300 to 1350 Notably in spite of the great
similarities with the various usages of the accent mark in Anglo-Saxon and
Norwegian written sources the exclusive and consequent employment of
the acute as a length mark is a unique feature of Old Icelandic script with
potential parallels only in Old Irish script (Lindblad 195210-11)
The reasons for this uniqueness may perhaps be sought in the role which
the First Grammatical Treatise is believed to have played in early Icelandic
scriptoria as a (or perhaps even the) normative text for apprentice scribes
until younger more influential fashions from continental Europe gradually
made their way to Iceland The anonymous author of the First Grammatical
Treatise recommends the use of a stroke for marking length (ldquomerkja ina
longu meeth stryki fraacute inum skommumrdquo 862-3 cf translation in Hreinn
Benediktsson 1972218-21) because he finds it more economic than the
solution employed by one of his models the Greek script which uses
separate graphemes for long vowels He is well aware of the fact that if he
were to devise a separate letter (stafr) for each of the units he is trying to
make order with he would get an alphabet of sixty-four units (thirty-six
vowels and twenty-eight consonants) Nonetheless he still wants to show
the distinction because ldquohon skiftir maacutelirdquo (862) Because of the lack of
parallels at the time (except maybe for Irish scriptoria) and possibly because
of the hint suggesting that the author had especially Greek in mind it is
sensible to assume that the usage of the accent mark was his original idea
just as the superscript dot denoting nasality among other innovations
The influence of the First Grammatical Treatise however if it ever was
that great does indeed fade away quite soon as Icelandic scribes become
29
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
more skilled and experienced Generally most of the early Old Icelandic
manuscripts make no graphic distinction between long and short vowels
because the traditional Latin orthography lacked this distinction as well as
for nasal and non-nasal vowels However the First Grammatical Treatise
testifies to a great level of linguistic awareness of an author or possibly a
small group of literate 12th-century Icelanders but since it was probably
written long before any extant manuscript it is hard to establish how great
its influence really was at the time It is for instance possible that a norm
like that of the length mark was but a recommendation which did not gain
widespread acceptance in other words marking vowel length was good
practice in those cases where not doing it would have made the reading
harder (which does only rarely occur)
In fact early manuscripts totally lacking length mark are not rare (cf
Lindblad 195228ff) AM 237 a fol for example dated to the mid-12th
century shows 5 times lteacutegt and 4 of the instances are the word verk which
has etymological short vowel and does not later diphthongise) Riacutembegla
(GKS 1812 4to A) has lteacutegt standing for ę just like ltoacutegt stands for ǫ On
the other hand there is a good number of sources where the length mark is
used selectively and according to an inner logic for example it may be
systematically avoided on prepositions adverbs and other non-nominal and
non-verbal monosyllables but nonetheless its notation elsewhere can be
extremely precise although the statistics will not account for that
It is therefore necessary in this present study to examine in details the
graphemic representation of subclasses VIIc and VIIf preterites in Old
Icelandic Within monographs dealing specifically with single manuscripts
and in other orthographic-statistical surveys it is not uncommon to
encounter statements of the kind ldquolteacutegt appears in N instances of
etymological short vowelrdquo where it is hard to understand whether the
30
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
author thinks that the preterites of subclasses VIIc and VIIf and related
cases originally had originally a short vowel or not
62 The earliest Icelandic Manuscripts Larsson (1981)
Already from a first glance into a reference work on Old Icelandic lexicon
in the earliest manuscripts it appears clear that where the length mark
occurs less often than it should when denoting length it is hardly used for
subclasses VIIc and VIIf preterites and almost never with the adverb heacuter
For a comparison the table presented below also shows the figures for the
commonest class VII verb heita It should also be mentioned that the
number of preterites which are attested in the first place is not as much as
desirable The following information is found in Larssons lexicon (1891)
including the following manuscripts AM 237 folio (abbr 237) Riacutembegla
GKS 1812 4to (abbr Rb) Icelandic Homily Book (Holm Perg 4to no15
abbr H) Physiologus Fragments (I II and III abbr Ph I II III in AM 673
4to A) AM 645 4to (abbr 645) Elucidarius (AM 674 4to A abbr El)
Placitusdraacutepa (AM 673 4to B abbr Pl) Graacutegaacutes (AM 315 folio D abbr
Grg)
31
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Table 7 Orthography of subclasses VIIc and VIIf strong preterites and heacuter in Larsson (1891)
237 Rb H Ph I Ph II Ph III 645 El Pl Grg
VIIa) HEITA 1sg heacutet3sg heacutet heacutet (29) het (5) het heacutet (44) het
(8) heacutet het (15) heiacutet
heacutet heacutett
het
3pl heacuteto (2) heto heacuteto (6)
VIIc) FAacute 1sg fecc fek fec2sg fect3sg feck (2) fek (7) fek fek(8) fec (4)
fexkfecc fecc
FALLA 3sg fell (12) fell (31) fell (2)
GANGA 1sg gek gek gec3sg geck (4)
gekc gecgeck (7) gek (8) geck
(2)gek (26) gec (16) gecſc
gek geck gek
HALDA 3sg helt helt (2) hellt (3) hellz
heacutelt helt (4) helt helt hellt
HANGA 3sg heck (3) hek (3) hek hec
VIIf) SAacute 3sg ſoslashre ſeacutere
SNUacuteA 3sg ſneore (2) ſneoreſc ſnoslashreſc (2)
ſnere (2) ſnereſc (2) ſneriſc
3pl ſnǿroſc ſnero (2) ſneroſc
Adv HEacuteR her her her heacuter
her (44) her (38) heacuter
her (2) her
her (3) her (16)
her (3) her (2) heacuter
her her (3)
In the whole corpus of manuscripts a subclass VIIc preterite is spelled only
once with lteacutegt (ltheacuteltgt in AM 645 4to) while the same word occurs four
other times in the same manuscript and in the same mood tense and
person with ltegt Class VIIa preterites on the other hand are
overwhelmingly spelled with lteacutegt as exemplified by heita the ratio
between ltheacutetgt and lthetgt is 678 against 1666 There is also one
6 The remaining spellings of ht (besides ltheiacutetgt) in AM 645 4to are abbreviated and are here believed to be less relevant for this kind of statistics
32
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
instance of a subclass VIIf preterite written with an accent mark
ltſnǿroſcgt which seems to be one of those cases as in ltheacuteltgt where the
accent mark may be superfluous given the very low attestation of subclass
VIIf preterites with accent mark Notably among all manuscripts there are
only three instances of ltheacutergt (of which one is actually lthgt in Holm Perg
15 4to see below)
63 Holm perg 15 4to
One of the most ancient and best known Old Icelandic manuscript is the
so-called Icelandic Homily Book (Homiliacuteuboacutek Holm perg 15 4deg) dated to
about 1200 Given its considerable size (102 pages) it represents an
extraordinary source of information about Old Icelandic language and
scribal practices It has been argued that if follows to some extent the
orthographical rules given by the First Grammatical Treatise to the extent
that even (albeit few) nasal vowels are marked (de Leeuw van Weenen
20044 and 60-61) There is traditionally little agreement on the number of
hands in the manuscript but the idea that the whole manuscript could have
been written either by one hand over a long period of time or by several
scribes all writing the same hand has been lately gaining more and more
popularity among the experts of this particular manuscript (de Leeuw van
Weenen 200434) In short thus the handwriting does vary considerably
but it is unclear whether this is to be traced back to the intervention of
several hands rather than to the will of being as close as possible even in
the layout to the parent manuscript(s) or more simply to a considerable
time span between the writing of its single parts
Hreinn Benediktsson (196733) wrote about the occurrence of the length
mark that it ranges from an average percentage of accuracy as high as 849
33
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
up to 968 for almost all hands with the only notable exception of the hand
called i exhibiting a meager 557 In fact differences within the
manuscripts can be substantial and spelling can vary considerably
According to Lindblad (195259-61) the cases where an accent mark is
placed on an indisputably short vowel are 150 which represents a frequency
of about 025-03 In the most recent edition of the ms de Leeuw van
Weenen (200458) estimates that the spelling ltegt for eacute surpasses lteacutegt
and other alternative spellings fill in the scenario as well as eacute occurs a few
times as ltęgt or ltaeliggt or lteigt Now these latter spellings are certainly
more valuable for the present research than plain ltegt as both ltaeliggt and
lteigt could perhaps say something about the early stage of diphthongisation
of eacute into [je] though [ej] [eⁱ] [aeligi] or a similar intermediate stage
The hand division showed in the table below is the one used by Andrea de
Leeuw van Weenen (2004) which is a sort of compromise among several
others and not differing too much from them As shown in the table
orthographical practices fluctuate considerably even within the same hands
Hand c in particular is inconsistent in the following points
a) consonant gemination k is written either ltkgt or ltckgt l(C)
either ltllgt ltlgt or ltgt
b) the spelling of the preterite of heita with 5 times ltegt vs 5 times
lteacutegt and this is about all of occurrences of lthetgt
c) most notably the spelling of subclass VIIf preterites
In this last case the normal root vowel is short oslash spelled either ltoslashgt or
lteogt but hand c spells ltſeacuteregt (one instance) which could suggest a long
derounded vowel and elsewhere the medio-passive ltſnoslashreſcgt while none
of subclass VIIc preterites bear an acute mark From the point of view of
historical phonology there can be no doubt that all cases of r-umlaut in this
34
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
kind of preterites had a short oslash and never a long one because its normal
development into modern Icelandic would have then been snaeligri
[snairɪ] rather than sneacuteri [snjerɪ] Hand q spells ltſnereſcgt but also
always ltheacutetgt without exceptions Given all other spellings elsewhere in the
manuscript it is therefore unlikely that ltſeacuteregt indicates a long vowel
Hand i is the only one marking a VIIc preterite but at the same time
spelling ltfellgt and lthetothorngt and it must therefore be deemed unreliable I
would thus draw the following conclusions about the manuscript
bull the normal vocalism in VIIc preterites is short e in VIIf short oslash
with some instances of derounding to e
bull the word here when not abbreviated appears nearly always as
lthergt (37 times) and only once as lthgt 73v22 (whereas Larsson
reported 44 see table above)
bull the vocalism of VIIa preterites is eacute as the acute mark surpasses by
far the instances where it is absent
bull it is possible that the spelling ltheacuteitogt 47v12 notes some kind of
diphthongisation of ē gt [ei] although it only occurs in one instance
35
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Table 10 Orthography of subclasses VIIa VIIc and VIIf strong preterites
divided by scribal hands in Hom perg 15 4deg as according to de Leeuw van
Weenen (2000)
Hand VII c VII f VII aFAacute FALLA GANGA HALDA HANGA SAacute SNUacuteA HEITA
a 1r 1v28-40 geb 1v1-27 heacutetc 2r-40v27 fek (3)
feck (2)fe fell geck (4) hellt (2)
helt heck (3) hek
ſeacutere ſnoslashreſc het (4) heacutet (5) heto
d 40v27-42ve 43r-44v fellf 45r-50v4 61v-
62v24geck heacutet heacuteito
g 50v5-54r heacuteth 54v-56vi 57r-61r feacutek fell hetothornk 62v25-65v18 fe (2) heacutet heacutetol 65v19-66v 69r ſoslashrem 67r-68vn 69v-77v fe gec he ſnoslashroſco 78r-80v3 ge (3) heacutet (3)p 80v4-35 94r19-
97rfecc fell (5) Ge ge het helt ſneoeſc het heacutet (11)
heacutetoq 81r-94r19 fe ge geck het he ſnereſc heacutet (8)r 97v-102v fell (2) ge (2) heacutetō
64 The earliest Norwegian Manuscripts Holtsmark (1955)
A comparison can be made within West Norse with the oldest Norwegian
manuscripts until ca 1250 The following table based on Holtsmark
(1955) contains the following manuscripts
bull Norwegian Homily Book I II III (AM 619 4to abbr Hom I II
III)
bull AM 655 IX 4to (abbr 655 IX)
bull Oacutelafs saga Tryggvasonar (in De la Gardie 4-7 fol abbr OT)
bull Norske Diplomer (in Norsk Riksarkiv abbr ND)
36
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Table 8 Orthography of of subclasses VIIc and VIIf strong preterites and heacuter in early Old Norwegian manuscripts according to Holtsmark (1955)
Hom I Hom II Hom III 655 IX OT ND I 51
VIIC) FAacute 3sg fecc (5) fec (7)
ecc (3)
FALLA 3sg fell (11) fel ell (4) aeligl ell (4)
GANGA 1sg gecc 3sg gecc (11)
gec (3)gec (3) gecc (4)
HALDA 2sg helz3sg helt (5)
hellthelt
HANGA 2sg hect3sg hecc (2) hec
VIIF) SAacute 3sg ſere (5)
SNUacuteA 2sg ſnereſc (2)3sg ſnere (2)
ſnereſc (2) ſnerez
ſnere ſneɼe
VIIA) HEITA 1sg3sg het (8)
heacutet (10)het (2) heacutet
heacutet hett het (2) hett
ADV HEacuteR her (19) haeligr (5)
her (25) heacuter haeligr
her (3) heɼ (4) her (2)
Two more manuscripts in Holtsmark (AM 315 G and Ra I C IV) read
ltheacutergt but with the caption ldquooverskriftrdquo (added later by an another hand
possibly the compiler) and therefore excluded here The overall situation
does not seem to be far from the early Old Icelandic subclass VIIa
preterites are sporadically marked whereas no VIIc and f preterites are
With 25 instances of lthergt for the adverb heacuter in Hom II and 19 in Hom I
against but one ltheacutergt in Hom II even this seems to be short statistically
given the great occurrence of the word ndash and the fact that it occurs so many
time in full ndash one would expect it to be closer to the values of heacutet at least in
AM 619 4to One could of course speculate that adverbs and other particles
received at times a different treatment from verbs nouns and other lexically
37
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
richer grammatical classes ndash but that is arguably much harder to claim than
when they are all but heavily abbreviated as they are elsewhere
65 NRA 52
Among the earliest known manuscripts are fragments of the saga of Olaf
the Saint (Oacutelafs saga helga hin elzta NRA 52) found in the Norsk Riksarkiv
and dated to ca 1225 In spite of its fragmentary state it is not hard to note
how the scribe took very good care in marking length The manuscript is
very accurately written also regarding the use of the accent mark denoting
length The following is the more recent evaluation by Lindblad (195288-
89)
[NRA 52] av samtliga fisl Hss har den naumlst houmlgsta accentfrekvensen vid laringng
vokal och diftong De 6 raumltt obetydliga fragmenten inneharingller ej mindre aumln 495
akutecken och laringng vokal accentueras i 77 av samtliga fall Det houmlgsta
percenttalet uppvisar AnR [Annales Regii] (88)
In the original portions of the ms most long vowels in class VII preterites
are systematically marked where they should be long (ie in all strong
preterites lttoacutekgt ltstoacutethorngt ltfoacutergt ltleacutetgt) and the scarce attestations of the
forms ltfekgt (one time) ltheltgt (one time) and ltgekgt (twice) points
towards a distinctively short vowel
66 GKS 2087 4to
The Annales or Konungsannaacutell (GKS 2087 4to) are also both very old and
precise among the sources which make use of the length mark The
38
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
manuscript was written partly in Latin and partly in Norse by two hands
one after another the older ending the chronicle in 1306 the younger
continuing until 1341 Unfortunately the instances which are relevant to
the present survey do not abound but even in their small number point at a
short vowel in VIIc preterites The following is a chart with all relevant VII
class verb forms in the manuscript based on Buergel (1904)
Table 9 Orthography of class VII strong preterites in GL kgl Sml 2087 4to
according to Buergel (1904)
Subclass Infinitive Person and Number
Orthography N of Occurrences
VIIa HEITA 3 sg (pres) heiacutet 23 sg heacutet 3
VIIb HLAUPA 3 sg liacuteop (hljoacutep) 4
VIIc FALLA 3 sg fell 13 pl fellv 3
HALDA 3 sg hellt 13 pl helldv 1
FAacute Inf faacute 13 sg fe 1
GANGA 3 sg ge 13 pl gengv
VIId LAacuteTA 3 sg leacutet 13 sg leacutetz 13 pl leacutetvz 2Part laacuteti 1Part laacutetn 1
RAacuteETHA 3 sg reacuteeth 13 sg reacuteethz 1
VIIf SNUacuteA 3 pl ſnervz 1
Here the root vowel is regularly marked in subclasses VIIa and VIId in
correspondence of an etymological long vowel on the other hand there is
no accent mark in subclass VIIc and in the only instance of subclass VIIf
39
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
67 AM 519a 4deg
The manuscript of Alexanders saga AM 519a 4deg dated ca 1280 consists of
the B version of the prose translation of the 12th century Latin epic poem
Alexandreis by Walter of Chacirctillon with a lacuna of four leaves within 37
parchment leaves The Old Norse translation is attributed to Brandr
Joacutensson (born ca 1210 died 1264) in AM 226 fol and Stock Perg 4deg nr 24
which are much younger than AM 519 a 4deg and contain the A version The
language is surely Icelandic and some traits point to an Icelandic scribe but
Norwegian influence is clear throughout the manuscript as was common in
Iceland at that time It is especially interesting for its paleography as it is
clearly written by a single hand and even the corrections seem to be written
by the same hand (or by a corrector using exactly the same ductus and
spelling) There are some occasional corrections and expansions by a 17th-
century hand (de Leeuw van Weenen 200925)
The length mark is used less regularly for single vowels and it never
appears on the preterites of faacute falla ganga halda and ganga (de Leeuw van
Weenen 2009141) As far as VIIf is concerned only the preterite of snuacutea is
attested 4 times appearing as either sneri or snoslashri (although in several
instances the ltegt is closed so that it is not really distinguishable from
ltoslashgt) The overall occurrences of lteacutegt are 94 in all cases against the 17524
for ltegt (de Leeuw van Weenen 200934) The phoneme e is spelled ltegt
79 times lteacutegt 60 times (de Leeuw van Weenen 200951) ltheacutetgt occurs only
4 times against the 43 times of unmarked lthetgt and lthergt occurs 12
times and never with the acute The scribe therefore only rarely uses the
accent mark to denote vowel length
40
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
68 AM 132 fol
The most interesting text for the present survey is doubtless Moumlethruvallaboacutek
(AM 132 fol) dated ca 1330-1370 The manuscript contains several sagas
included Laxdaeligla and Foacutestbraeligethra saga and was written mostly by one hand
as well as at least four minor medieval hands and a much later one from the
seventeenth century (de Leeuw van Weenen 200022)
Here according to Andrea de Leeuw van Weenen the verbs faacute falla ganga
and halda are well attested in their multiple preterite forms to the extent
that they exhibit parallel paradigms with alternation e~i (attested
secondary forms are noted in brackets all forms are normalised in the table
an their spelling discussed below)
Table 10 Variants of Subclass VIIc Preterites in AM 132 fol
faacute fekk (fengu) (fanginn)
feacutekk fingu fenginn
(finginn)
falla fell fellu fallinn
feacutell feacutellu
ganga gekk (gengu) gengit
gingu (gingit)
halda helt heldu haldinn
heacutelt heacuteldu
Although the spelling in Moumlethruvallaboacutek is generally remarkably precise the
notation of length is not to the extent that the editor states that ldquothe use of
accents in Moumlethruvallaboacutek is such that no conclusions for the length of
vowels can be drawn from itrdquo (de Leeuw van Weenen 200063) For words
other than VII class strong verbs only one instance of lteacutegt and two of ltiegt
41
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
are recorded for e eacute is usually spelled ltegt 2937 times lteacutegt 294 times
ltiegt 275 times ltiacuteegt 56 times lteigt 3 times ltieacutegt 1 time and ltiaeliggt 1 time
Lindblad (1952128) reports 557 occurrences of the accent mark with
ldquoetymologically long vowelrdquo out of the 576 single accents in the manuscript
Subclasses VIIc and VIIf strong preterites are clearly understood by
Lindblad to have a short vowel as it is mentioned that the accent mark
occurs but 15 times on etymologically short vowels of which once in feacutell
(23819)
But the most important innovation of the present manuscript is that a
diphthong [je] is written in the preterite singular of the verbs faacute falla and
halda According to the index by Andrea de Leeuw van Weenen (2000) the
preterite of faacute is written ltiekgt one time ltieckgt 9 times and 20 times
without diphthong while the preterite of falla appears 8 times as ltiellgt 2
times as ltiellugt 2 times as lteacutellgt with the accent mark and 34 times with
neither diphthong nor accent mark It seems therefore that the diphthong
[je] had already spread to the preterite plural of falla and halda The
preterite of halda appears as lthɩelltgt 3 times lthɩeltgt 2 times lthɩellugt 2
times and once as lthɩellɩgt for a total of 8 times as well as 2 times with
an accent mark as ltheacutelltgt against 40 occurrences with neither diphthong
nor accent mark
Class VII strong preterites however seem to be the only cases of
diphthongation in the manuscript since in addition to the preterites of
hanga (5 times) ganga (53 times) which are present in the manuscript but
do not show neither ltiegt nor accent mark other words which would be
expected to show ltiegt just like in feacutekk feacutell and heacutelt do not The adverb
here is written in full a number of times but always as lthergt (unlike
heraeth which is always but once abbreviated compounds included) forms
42
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
the verb eta with the present stem occur 14 times eacutel occurs once with an
accent mark
69 Summary
To summarise the evidence found in the earliest manuscripts of prosaic
texts is that the root vowel in subclasses VIIc VIIf and heacuter was originally
short until signs of diphthongisation appear in VIIc in the 14th century A
similar survey but of poetic texts is unfortunately impossible because the
largest part of the preterites here investigated are found in hyper-
characterised syllables which are not distinguished from the heavy syllables
in the metre (ie fekk (heavy) and feacutekk (hyper-characterised) would behave
in exactly the same way within the metre) Nonetheless a survey of some
subclass VIIf preterites and other instances of the sound change e gt [je]
has been carried out by Haukur THORNorgeirsson (mentioned in a talk given at
the University of Iceland in 2009 and personal communication) The scope
of Haukurs survey was poetry after 1200 until the time of the Quantity
Shift Out of 60 instances sneri was found 13 times and reri 2 times both
always with a short vowel This evidence clearly qualifies VIIf preterites as
the last group of all words discussed in the previous chapter to undergo
diphthongisation (according to Bjoumlrn Karel THORNoacuteroacutelfsson 1925115-116 as late
as the 18th century) These data also exclude the possibility that snoslashri was
short but the unrounded sneacuteri was long Even as late as the Guethbrandsbibliacutea
(1584) where VIIc preterites and related cases are now most often noted
with a diphthong (hiellt blies hiedan ietin) the spelling ltsneregt is the still
only one coming forth (cf Joacuten Helgason 192918 and Bandle 1956407) As
will be discussed in the final chapter it is likely that the spelling ltiegt
reflected some regional pronunciation which took relatively longer time to
43
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
spread across the island and be accepted into a now well-codified written
language
7 Conclusions
As we have seen both subclass VIIc (hellt fell gekk hekk etc) and VIIf
(sneri reri etc) preterites occur with a short vowel from their first
attestations in Old Icelandic until the middle of the 14th century when some
of the subclass VIIc preterites occur with the diphthong [je] in the preterite
singular (lthielltgt in Moumlethruvallaboacutek AM 132 fol from 1330-1370 cf
Chapter 7) Therefore the reconstruction of the Proto-Germanic forms has
to account for ĕ as the root vowel in Old Icelandic preterites The most
likely origin of the short root vowel is from the old reduplicating syllable
which would have come to bear the stress through the elision of the old
root vowel This could have either happened through ldquoejectionrdquo of the old
vowel (Jasanoff 2007) or more likely haplology of the entire old root
syllable (Joumlrundur Hilmarsson 1992) The reasons for preferring haplology
are chiefly two On one hand there is no evidence that a contraction ever
produced a radical ē which was then subsequently shortened in Old
Norse On the other hand a secondary (perhaps compensatory) lengthening
of a short e to ē2 is feasible for both subclasses VIIa (heacutet leacutek etc))
VIId (leacutet reacuteeth etc) and VIIe (bleacutet) which all have either a diphthong or a
long vowel in the present stem and an open syllable structure in the
preterite plural Concerning the archaic Anglian forms mentioned in
chapter 5 (eg heht reord) their retention of a part of the old root syllable
may well indicate that in Anglian reduplication was not yet altogether
obsolete this would mean that there is no reason to date the final loss of
reduplication back to Proto-Germanic times and that the development in
44
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Old Norse (and for sure Old Saxon) may have been independent from that
of Old English and Old High German
Presumably from a very early time a fluctuation in the vowel length of
several Old Icelandic words starts taking place Many of such words contain
e in the root It first affects (open) monosyllables which were liable of
being occasionally emphatically lengthened according to their position in
the sentence a sure example is the adverb her which is primarily noted
without accent mark Moreover as seen in chapter 5 ldquoarchiphonemesrdquo
primarily perceived as short where the correlation of quantity had been
neutralised are found in some instances of etymologically long vowels
preceding junctures and hiatus (Hreinn Benediktsson 1972138)
monosyllabic strong preterites such as vaacute (from vega) and fraacute (from fregna)
and in several forms of the present of all verba pura (included all verbs in
subclass VIIf eg snuacutea roacutea etc)
Summarising from the 14th century on the following sound changes take
place
a) long mid and open monophthongs were diphthongised included the
long root vowels in most class VII strong preterites (heacutet greacutet bleacutet
etc)
b) diphthongisation also occurred in vowels that had been interpreted as
short phonemically ie where the quantity correlation had been
neutralised before a juncture or a hiatus (snuacutea buacutea vaacute fraacute etc)
Diphthongisation also affected inherited short monophthongs because of
different processes
45
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
c) in words starting with h+e (heacuteethan heacuteraeth but also heacutekk and heacutelt in
subclass VIIc preterites)
d) the present root vowel of the class V strong verb eacuteta by analogy with
its preterite forms
e) shortened vowels in hyper-characterised syllables of neuter forms of
adjectives (skjoacutett fljoacutett etc) because of an ldquounderlying diphthongrdquo
extended from their masculine and feminine forms
Subclass VIIc preterites hekk and helt were probably the pioneers of the
diphthongisation to [je] because of their word-initial h Through this
process a first couple of preterites in subclass VIIc to be considered
typologically closer to the diphthongised ones in other subclasses the new
syllable structures CjeRC (hjelt) and CjeCC hjekk) are closer to
CjeC (hjet rjeeth) and CRjeC (grjet bljet) than the original
CeRC (helt) and CeCC (hekk) Gradually the analogical process spreads
across more forms of subclass VIIc preterites too but not to the plural of
forms containing the cluster ng whose root vowel had undergone another
diphthongisation to [ei] This is still evident in the modern pronunciation
[ej] of the preterite plural forms fengum and gengum (cf Sigfuacutes Bloumlndal
1924170 [fɛiɳgoslashm] and [gɛiɳgoslashm] which in mod IPA would be
[fejɳgʏm] [gejɳgʏm]) and also heacutengum which at times shows both
diphthongs ([hjejɳgʏm]) because of the word-initial h beside the regular
heacuteldum [hjeldʏm] and feacutellum [fjetlʏm] The diphthong [ei] in fengum and
gengum but also in heacutengum clearly go back to a short monophthong
Meanwhile the number of items in subclass VIIc is reduced as falda and
blanda become weak (cf Sigfuacutes Bloumlndal 1924172 and 84)
Lastly subclass VIIf preterites (snera rera etc) are affected by the
diphthongisation There are several reasons why the subclass VIIf preterites
46
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
got a diphthong much later than subclass VIIc preterites First as posited in
chapter 5 the opposition of quantity was neutralised in much of the present
stem possibly all the way to the Quantity Shift in the 16th century
Consequently the present could not be the source of analogical lengthening
of the root vowel in the preterite Secondly the bisyllabic structure of the
preterite singular forms on one hand and the dialectal fluctuation
between ouml and e which to a certain extent continues until as late as the
20th century (see below) contributed in keeping the subclass typologically
separated from the others Thirdly and as a consequence of the said
separation the verbs were soon reanalysed as of a new kind of weak type
as if it formed the preterite by means of an -r- infix for this reason the
second syllable in the preterite singular was perceived of as an inflected
ending comparable to that of a weak verb (snera gt sneri with ending -ri as
the weak dental suffix -ethi) Here too as late as the 20th century the process
is incomplete as the remaining four verbs in subclass VIIf all have non-
diphthongised variants at least until the 20th century Sigfuacutes Bloumlndal
writing in the 1920s reports the pronunciations [sn(j)ɛrɪ] and [sn(j)ɛroslashm]
with optional [j] and also optional pronunciation [snoumlrɪ] (Sigfuacutes Bloumlndal
1924769) [rɛrɪ] beside [rjɛrɪ] (and [roumlrɪ] in East Skaftafell East Fjords
and West Fjords) (1924657) similarly [grɛrɪ] [grjɛrɪ] [groumlrɪ] (1924274)
and [njɛrɪ] beside [noumlrɪ] in the East Fjords (1924583) Finally saacute (mod
pret saacuteethi according to the Beygingarlyacutesing iacuteslensks nuacutetiacutemamaacutels) migrated
to the weak proper conjugation forming the preterite with a dental suffix
47
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
8 Bibliography
Bammesberger Alfred 1994 Dehnstufe und Reduplikation im
Urgermanischen Bopp-Symposium 1992 der Humboldt-Universitaumlt zu
Berlin Akten der Konferenz vom 243 ndash 2631992 aus Anlaszlig von Franz
Bopps zweihundertjaumlhrigem Geburtstag am 1491991 Ed Reinhard
Sternemann pp 15-20 Carl Winter Heidelberg
Bandle Oscar 1956 Die Sprache der Guethbrandsbibliacutea Orthographie und
Laute Formen Munksgaringrd Hafniaelig
Beyingarlyacutesing iacuteslensks nuacutetiacutemamaacutels Ed Kristiacuten Bjarnadoacutettir Stofnun Aacuterna
Magnuacutessonar iacute iacuteslenskum fraeligethum Reykjaviacutek 2002-2010 Available at
httpbinarnastofnunis (accessed in May 2010)
Bjoumlrn Karel THORNoacuteroacutelfsson 1925 Um iacuteslenskar orethmyndir aacute 14 og 15 oumlld og
breytingar thorneirra uacuter fornmaacutelinu meeth viethauka um nyacutejungar iacute
orethmyndum aacute 16 oumlld og siacuteethar Fjelagsprentsmiethjan Reykjaviacutek
Bjoumlrn Karel THORNoacuteroacutelfsson 1929a Kvantitetsomvaeligltningen i islandsk Arkiv foumlr
Nordisk Filologi 4535-81
Bjoumlrn Karel THORNoacuteroacutelfsson 1929b Nokkur oreth um hinar iacuteslensku
hljoacuteethbreytingar eacute gt je og y yacute ey gt i iacute ei Studier tillaumlgnade Axel Kock
Lund CWK Gleerup Arkiv foumlr nordisk filologi tillaumlggsband till band
XL ny foumlljd 232ndash243
Braune Wilhelm 1967 Althochdeutsche Grammatik Niemeyer Tuumlbingen
48
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Bremmer Rolf H Jr 2000 An Introduction to Old Frisian History
Grammar Reader Glossary Benjamins Leiden and Philadelphia
Cleasby Richard and Gudbrand Vigfusson 1874 An Icelandic-English
Dictionary Clarendon Press Oxford
Fulk RD 1987 Reduplicating Verbs and their Development in Northwest
Germanic Beitraumlge zur Geschichte der Deutschen Sprache und Literatur
109159-178
Garnes Sara 1976 Quantity in Icelandic production and perception Buske
Hamburg
Haugen Einar 1982 Scandinavian Language Structures A Comparative
Historical Study Niemeyer Tuumlbingen
Haukur THORNorgeirsson 2009 Hin helgu eacute - stiklur uacuter erindi Handout of a talk
given at Haacuteskoacuteli Iacuteslands ndash University of Iceland on September 11th
2009
Heusler Andreas 1950 Altislaumlndisches Elementarbuch 3rd ed Carl Winter
Heidelberg
Holtsmark Anne 1955 Ordforraringdet i de eldste norske haringndskrifter til ca 1250
Gammelnorsk ordboksverk Oslo
Hreinn Bedediktsson 1959 The vowel system of Icelandic a survey of its
history Word 15 282-312
49
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Hreinn Benediktsson 1967 Indirect Changes of Phonological Structure
Nordic Vowel Quantity Acta Linguistica Hafniensia 1131-65
Hreinn Benediktsson (ed) 1972 The First Grammatical Treatise
Introduction Text Notes Translation Vocabulary Facsimiles Institute
of Nordic Linguistics Reykjaviacutek
Jasanoff Jay 2003 Hittite and the Indo-European Verb Oxford University
Press Oxford
Jasanoff Jay 2007 From Reduplication to Ablaut The Class VII Strong
Verbs of Northwest Germanic Historische Sprachfrschung 120241-284
Joacuten Helgason 1929 Maacutelieth aacute nyacuteja testamenti Odds Gottskaacutelkssonar Hieth
iacuteslenska fraeligethafjelagieth iacute Kaupmannahoumlfn Copenhagen
Joumlrundur Hilmarsson 1991 On ēsup2 in Germanic Acta Linguistica Hafnensia
2333-47
Katara Pekka 1939 Die urspruumlnglich reduplizierenden Verba im
Niederdeutschen Beitraumlge zur Geschichte der deutschen Verbalflexion
Meacutemoires de la Societeacute neacuteophilologique de Helsinki 12 Socieacuteteacute
neacuteophilologique Helsinki
Kristjaacuten Aacuternason 1980 Quantity in Historical Phonology Icelandic and
Related Cases Cambridge University Press Cambridge
50
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Kristjaacuten Aacuternason 2005 Hljoacuteeth Handboacutek um hljoacuteethfraeligethi og hljoacuteethkerfisfraeligethi
Iacuteslensk tunga I Almenna boacutekfeacutelagieth Reykjaviacutek
Larsson Ludvig 1891 Ordforraringdet I de aumllsta islaumlnska haringndskrifterna Ph
Lindstedts Universitets-bokhandel Lund
de Leeuw van Weenen Andrea 2000 A Grammar of Moumlethruvallaboacutek
Leiden University Press Leiden
de Leeuw van Weenen Andrea 2004 Lemmatized Index to the Icelandic
Homily Book Perg 15 4deg in the Royal Library Stockholm Stofnun Aacuterna
Magnuacutessonar aacute Iacuteslandi Reykjaviacutek
de Leeuw van Weenen Andrea 2009 Alexanders Saga AM 519a 4deg in the
Arnamagnaelign Collection Copenhagen Museum Tusculanum Press
Copenhagen
Lindblad Gustaf 1952 Det islaumlndska accenttaumlcknet En historisk-ortografisk
studie Gleerup Lund
Noreen Adolf 1913 Geschichte der nordischen Sprachen besonders in
altnordischer Zeit Truumlbner Straszligburg
Noreen Adolf 1923 Altnordische Grammatik I Altislaumlndische und
altnorwegische Grammatik (Laut- und Flexionslehre) unter
beruumlcksichtigung des Urnordischen 4th ed Niemeyer Tuumlbingen
Prokosch Eduard 1939 A Comparative Germanic Grammar Linguistic
Society of America Philadelphia
51
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Ringe Don 2007 From Proto-Indo-European to Proto-Germanic Oxford
University Press Oxford
Seip Didrik Arup 1954 Palaeligografi B Norge og Island Bonnier Stockholm
Sigfuacutes Bloumlndal 1923 Iacuteslensk-doumlnsk orethaboacutek (Reykjaviacutek 1920-1924) Hieth
iacuteslenska boacutekmenntafeacutelag Reykjaviacutek
Steller Walter 1928 Abriszlig der altfriesischen Grammatik Niemeyer Halle
(Saale)
Storm Gustav 1893 Otte brudstykker af Den aeligldste saga om Olav den hellige
Groslashndahl amp Soslashns Christiania
Sturtevant Albert Morey 1953 ldquoFurther Old Norse Secondary
Formationsrdquo Language Vol 29 No 4 457-462 The Linguistic
Society of America Washington DC
Torp Alf 1909 Woumlrterbuch der Indogermanischen Sprachen Dritter Teil
Wortschatz der Germanischen Spracheinheit Goumlttingen Vandehoek amp
Ruprecht
Van Coetsem Frans 1956 Das System der starken Verba und die
Periodisierung im aumllteren Germanischen Noord-hollandsche uitgievers-
maatschappij Amsterdam
52
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53
Vennemann Theo 1994 Zur Entwicklung der reduplizierenden Verben im
Germanischen Beitraumlge zur Geschichte der deutschen Sprache und
Literatur 116 Band Niemeyer Tuumlbingen
Voyles Joseph 1992 Reduplicating Verbs in North-West Germanic
Lingua 5289-123
53