Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Arabic stress in Strict CV, with no moras, no syllables, no feet and
no extrametricality
Noam Faust (Université Paris 8 / CNRS SFL), Shanti Ulfsbjorninn (UCL - London)
האב נישט קיין מורא!. 1
▪ Strict CV (Lowenstamm 1996, Scheer 2004) is a theory that deals primarily with V-ø
alternation. The basic assumption is that the only skeletal unit is the CV unit:
(1) Syncope in [katab, katbat] ‘he, she wrote’ in Strict CV
a. k a t a b b. k a t a b a t
│ │ │ ↓ │ │ │ │ │ │ │
C V1 C V2 C V3 C V1 C V2 C V3 C V
▪ Moraic Theory (e.g. Hayes 1995, Davis 2011) tried to do away with the skeleton.
Accordingly, in Strict CV moras are out of work.
▪ But how does one then account for all that’s been accounted for using moras??? First, don’t
panic. The only thing moras were ever indispensable for was metric weight.
▪ Scheer & Szigetvari (2005) (S&S) show that many stress-assignment systems that were
claimed to be weight-sensitive can be described as having immobile stress in Strict CV.
▪ For instance, the default stress system of Spanish (and many other languages)
C-final = final stress: [χamon] ‘ham’
V-final = penultimate stress: [paloma] ‘pigeon’
…can in Strict CV be described as simply stressing the penultimate V.
▪ However, as shown in Ulfsbjorninn (2014), there are still weight-sensitive languages in
which stress-assignment cannot be analyzed as an edge-effect, languages in which (non-edge)
CVC and CVV can be shown to attract stress from its otherwise expected position.
▪ If one is to hold that moras don’t exist, one must account for such systems; this is what I will
do today with that bastion of Moraic Theory, stress in Arabic dialects. I will focus on
Palestinian and Cairene.1
▪ The analysis will be supported by the ease with which it can be extended to the issue of the
distribution of vowel length in the language, a topic with which Moraic Theory would
struggle.
1 Yoshida (1993) includes an appendix with a preliminary account of Palestinian stress in Strict CV, in which
stress is immobile: the algorithm is simply “stress the antepenultimate nucleus”, counting the FEN. But this
makes the wrong prediction for words like /sakkatat/ ‘she silenced’, which are stressed [sakkatat], not
*[sakkatat].
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
2
2. Strict CV Metrics and syllable weight in Palestinian and Cairene Arabic
(2) Basic stress facts of Palestinian Arabic (Brame 1974, Hayes 1995, Watson 2011)
a. (ante)penultimate in LL(L) nisi
kátab-u
‘he forgot’
‘they wrote’
b. stress to weight I biħibb
bisˁiːr
katab-t
‘he loves’
‘he becomes’
‘I wrote’
c. stress to weight II tarʒam-ti
jisaːfir
jitarʒim
‘you (fm) translated’
‘he should travel’
‘he should translate’
d. window effect sakkatat ‘she silenced’
sakkatato ‘she silenced him’
(3) Differences between Palestinian and Cairene
‘desk’ ‘she silenced’ ‘they travelled’
a. Palestinian maktabe sakkatat saːfaru
c. Cairene maktaba sakkatat saːfaru
▪ First, the notion of weight has to be integrated into a Strict CV model. One way to do so was
put forth in Ulfsbjorninn (2014).
(4) [katabt] ‘I wrote’ using the notions of projection and incorporation
a. Projection (before Incorporation) b. after incorporation
L3 L3 *α
L2 * * L2 * *
L1 * * * L1 * * *)α
k a t a b t k a t a b t
│ │ │ │ │ │ │ │ │ │ │ │
C V1 C V2 C V3 C V C V1 C V2 C V3 C V
(5) Palestinian Stress assignment2
a. Define window among last three Line 1 projections.
b. Project EN Line 1, CN to Line 2, do not project FN.3
c. Incorporate EN to preceding CN
d. If L3 is not reached, project the antepenultimate MSN (metrically-significant
nucleus) to L3 and stress L3 projection.
e. If that is an empty nucleus, stress the preceding MSN.
2 EN=empty nuclei; CN=contentful nuclei; FEN = Final Empty Nucleus.
3 Not projecting the FEN derives final-C extrametricality (Charette 1984).
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
3
(6) More examples
a. L3 *α d. L3 *
L2 * * * L2 * * *
L1 * * *α) * L1 * * *
k a t a b t i k a t a b u
│ │ │ │ │ │ │ │ │ │ │ │ │
C V C V C V C V C V C V C V
c. L3 *α d. L3 *α
L2 * * * L2 * * * L1 * *α) * * L1 * *α) * * m a k t a b e s a f a r u │ │ │ │ │ │ │ │ │ │ │ │ │
C V C V C V C V C V C V C V C V
(7) Cairene: same algorithm, except that stress is protracted if L3 is empty
a. L3 *α * b. L3 *α *
L2 * * * L2 * * *
L1 * *α) * * L1 * *α) * *
m a k t a b a s a k a t a t
│ │ │ │ │ │ │ │ │ │ │ │ │ │
C V C V C V C V C V C V C V C V C V
(8) Shifts or lack thereof in cliticized words
a. Palestinian V=Object clitic
sakkat-at ‘she silenced’ sakkat-at=o ‘she silenced him’
kaːtab-at ‘she corresponded’ kaːtab-at=o ‘she corresponded with him’
b. Palestinian N=possessive clitic
barak-e ‘blessing’ barak-it=na ‘our blessing’
barak-t=o ‘his blessing’ (*barakito)
c. Cairene V=Object clitic
katab-et ‘she wrote’ katab-et=u ‘she wrote it’
▪ It cannot be a coincidence that all of these exceptions occur in cliticized words…
(9) Stress clash and retraction in Palestinian cliticized nouns
a. L3 *
L2 * * *
L1 * * *
b a r a k e
│ │ │ │ │ │
C V C V C V
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
4
b. c. L4 *
L3 * * L3 *
L2 * * * * L2 * * * *
L1 * * * * L1 * * * *
b a r a k i t o b a r a k i t o
│ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │
C V C V C V C V C V C V C V C V
▪ Prediction for Cairene: when clash arises, stress should be advanced rather than retracted, in
order to stay within the window, just as in the /mak_taba/ => [maktaba] case.
(10) Stress clash and protraction in Cairene cliticized verbs
a. L3 *
L2 * * *
L1 * * *
k a t a b e t
│ │ │ │ │ │ │
C V C V C V C V
b. L3 * * c. L3 * *
L2 * * * * L2 * * * *
L1 * * * * L1 * * * *
k a t a b e t o k a t a b e t o
│ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │
C V C V C V C V C V C V C V C V
▪ The account also covers the window effect of Palestinian:
(11) No stress clash and no stress shift in Palestinian cliticized verbs: [sákkatat] vs.
[sakkátato]
a. L3 *α
L2 * * *
L1 * *α) * *
s a k a t a t
│ │ │ │ │ │ │
C V C V C V C V C V
b. L3 *α *
L2 * * * *
L1 * *α) * * *
s a k a t a t o
│ │ │ │ │ │ │ │
C V C V C V C V C V
3. Comparison to Moraic Account (Hayes 1995)
▪ Hayes (1995): all nuclei project moras. Trochaic feet are then built on top of those moras
from left to right.
(12) [{niμsiμ}] = one foot.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
5
▪ But [{kaμtaμ}buμ] = also one foot…
▪ The rightmost syllable and mora remain unparsed, because there is a ban on non-binary feet.
▪ In our account, in contrast, the first MSN of the window is stressed in both [nisi] and
[katabu], and nothing else has to be added. Unlike Hayes’s account, our account does not
require the distinction between parsed and unparsed entities of any sort.
▪ For /biħibb/ and /bisˁiːr/ (and also for /katab-t/ ‘I wrote’), building feet on top of projected
moras gives the wrong prediction:
(13) *[{biμħiμ}bb] *[{biμsˁiμ}ːr] *[{kaμtaμ}bt]
▪ Unless feet are in fact not built directly on moras,
/bisˁiːr/ is syllabified as [(biμ)σ(sˁiμːμ)σr].
…and two heterosyllabic moras cannot be grouped together under the same foot:
(14) *[{(biμ)σ(sˁiμ}ːμ)σr].
▪ Yet this does not immediately solve the problem of *[biħibb]: all other things being equal,
there are two moras in this word [biμħiμbb].
▪ To explain why /ħib/ behaves like /sˁiː/, moraic theory makes the assumption that that coda
consonants also project a mora. If this is true, then *[{(biμ)(ħiμ}bμ)b] is also impossible.4
(15) Proposed parses according to Hayes (ignoring the syllabic level):
[biμ{ħiμbμ}b] [biμ{sˁiμːμ}r] [ka{taμbμ}t]
(16) Problems:
a. Yet another type of unparsed entity, the initial syllable. No unparsedness in our
account.
b. Crucially relies on syllables and syllabification (only case in the moraic account that
does). Syllables are absent from our account.
c. Two very different entities are relevant for the computation of stress: V, (internal)
coda. Only V in our account.
▪ It is unclear why a coda (and moreover, only an internal one; see below) should add to the
stressability of a preceding vowel. Moraic codas are a way to encode this, but they do little
more than formalize the facts, as opposed to explaining them.
▪ We contend that the heaviness of CVC and CVV in the moraic account is equal by mere
assumption, whereas in our account it follows from the general assumptions of CVCV
Phonology and the mechanism of Incorporation.
4 The same can be shown for [katabt].
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
6
▪ In addition, consider the word [tarʒam-t]. Under the moraic view, two feet will be created:
(17) [{taμrμ}{ʒaμmμ}-t].
▪ Hayes claims that when there are two feet, the rightmost of the two bears main stress. Yet
this cannot be the entire story.
▪ If i. coda consonants project a mora, and ii. the rightmost foot is stressed, then one wrongly
predicts
(18) *[jiμ{saμːμ}{fiμrμ}] *[{taμrμ}{ʒaμmμ}].
▪ Proposed solution: final consonants are extrametrical (even though they are codas):
(19) [jiμ{saμːμ}fiμ<r>] [{taμrμ}ʒaμ<m>] [{kaμtaμ}<b>].
▪ But why is it possible for a final coda not to project a mora, while internal codas must
project a mora, if both are codas? The fact that a consonant is situated at the right edge should
not be of any consequence.
▪ Final consonant extrametricality is a formalized observation, rather than an explanation.
▪ In contrast, our account manipulates a fundamental and independently necessary distinction
in the theory between FENs and governed ENs: only the former may remain empty even
though they are not governed. Our account does not require any unmotivated stipulation
about extrametricality.
▪ Hayes’s account as it was presented until now does not work well for HLL words like
[madrase], which is predicted to be stressed on the rightmost foot:
(20) *[{maμdμ}{raμseμ}]
▪ To explain these forms, Hayes proposes another type of extrametricality, which targets feet:
if a foot is at the right edge of the word, it will be ignored by the rule that stresses the
rightmost foot.
▪ But rightmost foot extrametricality leads to yet another complication. Consider a word like
[tarʒam-t], which was parsed [{taμrμ}{ʒaμmμ}-<t>]: the rightmost foot is not extrametrical in
this case. Hayes proposes that this is so because it is not really at the right edge.
▪ But if so, the footing of [sakkatat] also results in a foot that is not adjacent to the right edge,
and stress is wrongly predicted to be penultimate:
(21) *[{saμkμ}{kaμtaμ}-<t>]
▪ In order to explain the difference between [sakkatat] and [tarʒamt], Hayes stipulates that a
final consonant is “syllabified” with a preceding CV sequence (22a), but not with a preceding
CVC sequence (22b).
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
7
(22) Final C in the moraic analysis of Palestinian stress in Hayes (1995)
a. F < F > b. F F
│ │ │ │
σ σ σ σ σ
│ │ │ │ │
μ μ μ μ μ μ μ μ
│ │ │ │ │ │ │ │
s a k k a t a <t> t a r ʒ a m <t>
▪ Our account does not require the distinction between syllabified and unsyllabified entities of
any sort.
▪ The parallel to foot extrametricality in out account was the mechanism of stress retraction.
However, unlike foot extrametricality, which is a trick, it is a consequence of the entire
analysis (it occurs because the first position in the window is empty).
▪ There is also one aspect of Hayes (1995) which might at first glance support it over our
view: Hayes proposes that the only difference between Palestinian and Cairene is final foot
extrametricality, which Cairene simply lacks.
▪ This indeed derives the correct result for [{maμdμ}{raμsaμ}], as well as stress shift in Cairene
[{kaμtaμ}{beμtoμ}] vs. [{kaμtaμ}beμt] and the other forms which we analyzed as bi-cyclic.
▪ But in fact, our notion of stress retraction is equally elegant. Indeed, just like Hayes, we
proposed a minimal difference between the two dialects: Palestinian retracts, Cairene
protracts.
▪ A Final disadvantage that our account may be claimed to carry is its use of a three MSN
window, which was crucial in cases like [sakkatato].
To conclude the comparison: Hayes’s account involves no less than five types of ignored
constituents: (i) consonant extrametricality;
(ii) foot extrametricality;
(iii) unparsed syllables, initial and final;
(iv) unparsed moras;
(v) unsyllabified consonants.
In contrast, our account does not require any part of the representation to be unparsed,
unsyllabified or extrametrical. The only entity that is ignored in our account is the FEN, as a
result of an independently-motivated parameter setting.
▪ Even more nicely, our account easily extends to the issue of final vowels in Palestinian.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
8
3. Shortening in Palestinian
3.1. Data from Palestinian
▪ Long vowels shorten before heavy syllables (CVV: 22a.iii., CVC:22b.iii, 22d.ii).
▪ But long vowels do not shorten if they are
=> in closed syllables (22b.ii)
=> separated from the heavy syllable by a light syllable (22c.ii)
=> before a non-heavy syllable, even when the latter is stressed (22.d.iii).
(22bis) Metrical Vowel shortening in Palestinian (Abu-Salim 1986)
a. i. baːb ‘door’ c. i. moːlad ‘birthday’
ii. baːb=o ‘his door’ ii. moːlad-eːn ‘two birthdays’
iii. bab-eːn ‘two doors’ d. i. kaːtab-at ‘she corresponded’
b. i. makaːtib ‘offices’ ii. katab-t ‘I corresponded’
ii. makaːtb-eːn5 ‘our offices’ iii. kaːtabat=o ‘she corresponded with him’
iii. makatib=na ‘our offices’
▪ Abu Salim’s is the only analysis of this data set. We will show that our account is not only
superior to Abu-Salim’s, but also makes correct predictions regarding the longstanding issue
of:
(23) Final vowel shortening in Palestinian (e.g. McCarthy 2005 for Egyptian)
a. katab-u ‘they wrote’ c. ʔirmi ‘throw!’
katab-uː-li ‘they wrote to me’ ʔirmiː-ha ‘throw her!’
b. nisi ‘he forgot’ d. ħabb-ha ‘he loved her’
nisiː-t ‘I forgot’ ħabb-haː-ʃ ‘he didn’t love her’
3.2. Clash-induced shortening
▪ Essentially Moraic account (Abu-Salim 1986): the technicalities of stress assignment are
not really explained in the paper. The representations are reproduced here.
(24) Node preceding H is labeled “weak”, long vowels cannot survive under “weak” label
a. /baːb-eːn/ => [babeːn] b. /makaːtib-eːn/ => [makaːtbeːn]
| |
s w s
│ |
w s w s w s w │ │ | │ │ │ | b aμ aμ b eμ eμ nμ m a k aμ aμ tμ b eμ eμ nμ
5 This form, like all cases of long vowels in non-final closed syllables, involves syncope of the base’s short /i/.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
9
c. /moːlad-eːn/=>[moːladeːn] d. /kaːtab-at=o/=>*[kátabato]
s
│
s
| │
w s s
│ | │
s w s w w s w w │ │ │ | │ │ │ │ m oμ oμ l aμ d eμ eμ n k aμ aμ t aμ b aμ t o
▪ Abu-Salim is at a loss with (24d). He produces a story that is so problematic that I will not
repeat it here, and which he moreover abandons later and opts for a very shaky claim of
homophony avoidance: the expected *[katabato] is in fact identical to the underlying
/katabato/ ‘she wrote it’, which undergoes syncope and is realized [katbato]. By analogy
(between a surface form and a UR?) a shortened /katabato/ would also be expected to be
realized [katbato]. In order to avoid this, it is not shortened.
▪ We consider instead that the lack of shortening in [kaːtabato] indicates instead that Abu-
Salim’s account is simply wrong (not to mention the obscure and unorthodox footing
methods!).
▪ The main insight of Abu-Salim’s account is that /baːbeːn/ ‘two doors’ is problematic
because it involves equal prominence. Crucially ‒ and this is ignored by Abu-Salim ‒ the
problem of equal prominence is determined without reference to main stress.
(25) Vowel shortening in present approach: the problematic form as clash
L3 *α *β
L2 * *
L1 * *)α * *)β
b a b e n
│ │ │ │ │
C V1 C V2 C V3 C V C V
We claim that clash in (25) is avoided by undoing the incorporation of V2.6
▪ This undoing has a segmental effect: vowels can only become long by spreading into an
incorporated position. Since the position is not incorporated metrically, it cannot be identified
by segmental spreading. Because the vowel does not spread to V2, the result will be a short
vowel associated only to V1.
6 Here we opt for undoing incorporation and this leads to shortening, rather than directly shortening the vowel.
This analytical choice is confirmed in section 3.2.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
10
(26) Vowel shortening: Clash, no incorporation, no spreading into unincorporated position
L3 *β
L2 * *
L1 * * * *)β
b a b e n
│ │ │ │ │
C V1 C V2 C V3 C V C V
(27) No vowel shortening in closed syllables: double incorporation, no clash
L4 *β
L3 *α *γ
L2 * *
L1 * * *)α *)β * *) γ
m a k a t b e n
│ │ │ │ │ │ │ │
C V1 C V2 C V3 C V4 C V5 C V C V
▪ From a traditional syllable perspective this would be illogical. If a CVV syllable shortens
before CVV, then CVVC (which is heavier still) should have more incentive to shorten – not
less.
▪ For [moːladeːn], no clash is expected because the two incorporating nuclei (V2 and V5) are
separated, at the level of their heads (Line 2) by a (filled) unincorporated nucleus (V4):
(28) Vowel shortening in Palestinian: no clash in [moːladeːn]
L3 *α *γ
L2 * * *
L1 * *)α * * *) γ
m o l a d e n
│ │ │ │ │ │ │
C V2 C V3 C V4 C V5 C V C V
▪ Meanwhile, for [kaːtabato], which was a problem for Abu-Salim, no clash is expected
between V3, which will eventually bear stress, and V1, since clash is crucially defined
between local incorporating heads at Line 3 before main stress assignment:
(29) Vowel shortening in Palestinian: no clash in [kaːtabato]
L3 *α
L2 * * * *
L1 * *)α * * *
k a t a b a t o
│ │ │ │ │ │ │ │
C V1 C V2 C V3 C V4 C V
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
11
▪ Finally, consider forms like /bint-eːn/ ‘two girls’, derived from /bint/ ‘girl’ and realized
[binteːn] (the unsuffixed base is sometimes realized with epenthesis [binit]). These are not
discussed by Abu-Salim, but are relevant.
(30) /bint-eːn/ in both accounts
▪ Problem: no segmental effect of the
bimoraic syllable in weak branch. Abu-
Salim must limit the effect to vocalic
bimoraic syllables by assumption.
a.
w s
│ │
b iμ nμ t eμ eμ <n>
b. L3 *β ▪ Incorporation undone, but no
segmental effect is expected,
since no spreading is
involved.
L2 * *
L1 * * * *β)
b i n t e n
│ │ │ │ │ │
C V1 C V2 C V3 C V4 C V
To summarize, our account does not involve
1) funny homophony avoidance.
2) recursive feet etc.
3) the specification of the effect to strictly vocalic monosyllabic feet.
…and it also has another advantage, namely that it explains…
3.3. Position-induced shortening
(31) Final vowel shortening in Palestinian
a. katab-u ‘they wrote’ c. ʔirmi ‘throw!’
katab-uː-li ‘they wrote to me’ ʔirmiː-ha ‘throw her!’
b. nisi ‘he forgot’ d. ħabb-ha ‘he loved her’
nisiː-t ‘I forgot’ ħabb-haː-ʃ ‘he didn’t love her’
▪ McCarthy (2005) concludes that (i) underlyingly, all final vowels are long; and (ii) the
phonological computation shortens them.
▪ We accept (i) without discussion; (ii) then follows from our account above in the following
way:
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
12
(32) FEN (V4) unprojected, unincorporated and unable to host spreading
a. L3 *
L2 * * *
L1 * * *
k a t a b u
│ │ │ │ │ │
C V1 C V2 C V3 C V4
(33) Old FEN (V4) projected, incorporated and able to host spreading
b. L3 *β
L2 * * * *
L1 * * * *β) *
k a t a b u l i
│ │ │ │ │ │ │ │
C V1 C V2 C V3 C V4 C V3 C V4
▪ The Incorporation and demotion account of clash-induced vowel shortening developed in
the previous subsection easily extends to final vowel shortening,
▪ This fact lends independent support to
1) our explanation for metrical vowel shortening, and
2) our account of basic stress, in which FENs were unprojected.
To summarize, this has been a unified account of Palestinian Arabic
(i) stress
- Stress antepenultimate MSN unless Line 3 attained
(ii) Metrical (clash-induced) shortening
- Resolve clash between adjacent Line 3 incorporating heads
(iii) Positional (=final) Shortening
- no incorporation of unprojected MSN
4. Beyond Palestinian
▪ The natural next step would be to see if this kind of clash-based long vowel shortening is
attested in other languages, and if so, whether the conditions for shortening are the same.
(34) Slovak shortens the 2nd
of two adjacent long vowels (Kenstowicz & Kisseberth 1979)
Suffix with V: Multiple suffixes with V:
3P.S /-aː/ Freq. /aːv/ + 3P.S /-aː/
Root with V vol vol-aː vol-aːv-a ‘call’
Root with V: tʃiːt tʃiːt-a tʃiːt-av-a ‘read’
(We will not discuss it further, however, because the specific dialectologicial and syllable
structure details would require an excursus too long for this presentation.)
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
13
▪ Gidabal (Kenstowicz & Kisseberth 1979:180) differs in a simple and systematic from
Palestinian.
(35) Gidabal vowel shortening
(a) second of two adjacent vowel shortens
/badi + yaː/ badiyaː ‘should hit’
/yagaː + yaː/ yaga:ya ‘should be fixing’
cf. /gaːda + ya:/ gaːdayaː ‘should chase’
(b) vowel shortening also applies across codas (=empty nuclei)
/mani + yaːgan/ maniyaːgan ‘to get wallaby’
/muru:n + yaːgan/ muruːndagan ‘to get firewood’7
/gadi + beː/ gadibeː ‘right here’
/buru:r + beː/ buru:rbe ‘only two’
/njule + daːŋ/ njuledaːŋ ‘he (emphatic)’
/yu + daːŋ/ yudaːŋ ‘much later’
/nu:n + daːŋ/ nuːn daŋ ‘too hot’
▪ This major difference with Palestinian is systematically expected in our framework iff
empty nuclei in Gidabal are not incorporated, because only in that cases CVVC and CVVs
would be at the same height: L3.
(36) /muruːn + yaːgan/ > [muruːndagan] ‘to get firewood’
L4
L3 *α *γ
L2 * *
L1 * * *)α * *) γ
m u r u n d a g a…
│ │ │ │ │ │ │ │ │
C V1 C V2 C V3 C V4 C V5 C V C V
▪ That governed EN are not MSU is in fact empirically justified: stress in Gidabal falls on the
first long vowel, or if there are none on the first vowel (Geytenbeek & Geytenbeek 1971);
Internal codas do not make the syllable heavy.
▪ To be sure, as in Palestinian, clash in Gidabal still only affects local incorporating heads:
non-adjacent long vowels are permitted if they’re interpolated by a filled nucleus.
7 /y/ is transformed into a homorganic plosive after a nasal.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
14
(37) /gaːda + yaː/ [gaːdayaː] ‘should chase’
L3 *α *γ
L2 * * *
L1 * *)α * * *) γ
g a d a y
│ │ │ │ │ │
C V2 C V3 C V4 C V5 C V
▪ Encouraging results! Further studies to come…
5. Final words
▪ It is possible to account for weight-sensitive systems without moras (feet, syllables,
extrametricality, extrasyllabicity, unparsedness etc.)
▪ Moreover, this kind of effort can lead to insights on other prosodically-related phenomena in
the language, such as long vowel distribution (in the case of Palestinian).
▪ Phonology is immoral.
References
Brame, M. 1974. The cycle in phonology: Stress in Palestinian, Maltese, and Spanish.
Linguistic Inquiry 5. 39-60.
Charette, M. 1984. The Appendix in Parametric Phonology. Studies in African Linguistics.
Supplement, 9:49–53.
Charette, M. 2008. The vital role of the trochaic foot in explaining Turkish word endings.
Lingua, 118: 46-65.
Davis, S. 2011. Quantity. In John Goldsmith, Jason Riggle and Alan Yu (eds.), The Handbook
of Phonological Theory (2nd edition). pp. 103-140, Oxford: Wiley-Blackwell.
Geytenbeek, Brian & Helen.1971. Gidabal grammar and dictionary. Australian Institute of
Aboriginal Studies 43. Canberra.
Hayes, B. 1995. Metrical Stress Theory. Chicago: University of Chicago Press.
Kenstowicz, Michael & Charles Kisseberth. 1977. Topics in phonological theory. New York:
Academic Press.
Lowenstamm, J. 1996. CV as the Only Syllable Type. In: Durand, J. & B. Laks (eds.),
Current Trends in Phonology Models and Methods. European Studies Research
Institute, University of Salford. 419-442.
McCarthy, J.J. 2005. The length of stem-final vowels in Colloquial Arabic. In Alhawary,
Mohammad T. and Elabbas Benmamoun (eds.), Perspectives on Arabic Linguistics XVII-
XVIII. Amsterdam: Benjamins. 1-26.
Scheer, T. 2004. A Lateral Theory of Phonology. Vol 1: What is CVCV, and why should it be?
Berlin: Mouton de Gruyter.
Scheer, T. and P. Szigetvari. 2005. Unified representations for stress and the syllable.
Phonology, 22(1):37–75.
Faust & Ulfsbjorninn Tel Aviv university Phonology Circle
Arabic Stress with no mora March 11th
2018
15
Ulfsbjorninn, S. 2014. A Field Theory of Stress: the role of empty nuclei in stress systems.
Ph.D. Dissertation. SOAS – University of London.
Yoshida, S. 1993. Licensing of empty Nuclei: the case of Palestinian vowel harmony. The
Linguistic Review, 10:127-159.
Watson, J. 2011. Word stress in Arabic. In: van Oostendorp, M., Ewen, M., Hume, E., and K.
Rice (eds.), The Blackwell companion to phonology. Wiley-Blackwell, Oxford.