Upload
marie-langer
View
218
Download
0
Embed Size (px)
Citation preview
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 1/318
LINGUISTICS AND FORMULAS IN HOMER
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 2/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 3/318
LINGUISTICSAND FORMULAS
IN HOMERScalarity and the description of the particle per
by
EGBERT J. BAKKER
JOHN BENJAMINS PUBLISHING COMPANYAMSTERDAM/PHILADELPHIA
1988
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 4/318
Library of Congress Cataloging in Publication Data
Bakker, Egbert J.Linguistics and formulas in Homer: scalarity and the description of the particle per / by
Egbert J. Bakker.
p. cm.Bibliography: p.Includes indexes.1. Homer -- Language - Grammar. 2. Greek language — Particles. 3. Per (The Greekword). 4. Oral-formulaic analysis. I. Title.PA4201.B35 1988883' .01- dc l9 88-10110ISBN 90 272 2057 3 (Eur.) / ISBN 1-55619-046-8 (US) (hb. ; alk. paper) CIP
® Copyright 1988 - John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or
any other means, without written permission from the publisher.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 5/318
TABLE OF CONTENTS
Preface ix
1 INTRODUCTION 11.0 Aims of the investigation 1
1.1 Per in Homer and after 11.2 Facts from the data-base 31.3 Per in Greek linguistics 7
1.3.1 Emphasis 71.3.2 Concession and strengthening 91.3.3 Fraenkel: "per heisst 'auch'" 111.3.4 Etymology and diachrony 12
1.4 Categories and prototypes 141.5 Linguistics and formulas in Homer 19
NOTES TO CHAPTER 1 232 THE LINGUISTIC BACKGROUND: SCALARTTY 27
2.0 Introduction 272.1 Scales 28
2.1.1 Scalarity: introductory 282.1.2 Superlatives: scalar and attributive 302.1.3 The non-referentiality of scalar superlatives 312.1.4 Irrealis modality 32
2.2 Scalarity and referentiality 362.2.1
Attributive and referential superlatives 362.2.2 Intension and extension 372.2.3 Scalarity and extensional terms 39
2.3 Focus particles and focus contexts 402.3.1 Even as a focus particle 40
2.3.1.1 Conventional implicature and the meaning of even . .402.3.1.2 Elocution and the use of even 432.3.1.3 Focus and scope 442.3.1.4 Scope dependence and scope independence 452.3.1.5 Scope ambiguity 46
2.3.1.6 Even if: concession and concessive conditionals . . . .48
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 6/318
vi CONTENTS
2.3.2 Types of focus context 492.4 Negative and positive polarity 522.4.1 Positive polarity items 522.4.2 Negative polarity items and scale reversal 54
NOTES TO CHAPTER2 57
3 SCΑLΑRITΎ Α Ν D ΤΗ Ε Ρ RTICLE PER 673.0 Introduction 67
3.1 Intensional terms 683.1.1 Scalar superlatives 683.1.2 Scalar superlatives and the etymology of per 713.1.3 Epic te and generic modality 723.1.4 A remark on diachrony 73
3.2 Extensional terms 753.2.1 The intension of focus constituents 763.2.2 Superlativeness, comparativeness and the illocution of even . 773.2.3 The concessive relative clause 803.2.4 Scalar expressions and the Homeric narrator 81
3.3 Scalarity and comparatives 833.4 Negative polar scalarity 85
3.4.1 Declarative sentences 853.4.2 Non-assertive contexts and negative polarity 863.4.3 Scalar superlatives and negative polarity 873.4.4 The strengthening of negation 89
3.5 Scalarity in wishes and commands 903.5.1 Imperative and optative sentences 903.5.2 Inclusiveness and exclusiveness 91
3.5.3 Scales of desirability and acceptability 943.5.4 Per,ge and exclusion 973.5.5 The diachronic aspect of per in wishes 98
3.6 Some unclassifiable instances 98
NOTES TO CHAPTER 3 101
4 PARTICIPLES I: LINGUISTICS 1074.0 Introduction 1074.1 On participles 1084.1.1 Kinds of adverbial relation 108
4.1.2 Syntactic independence and modal autonomy 1104.2 From scalarity to concession 113
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 7/318
CONTENTS vii
4.2.1 The intension of circumstances 1134.2.2 Diachronic aspects of kai hôs 1164.2.3 Syntactic intermezzo: correlation and apposition 117
4.3 The concessive participial phrase 1204.3.1 The prototypical properties of participle + per' 1204.3.2 Loss of superlativeness 1244.3.3 Dependent participles 1314.3.4 Modally embedded participles 133
4.4 From linguistics to the formula 1374.4.1'Participle + pef and generic modality 1374.4.2 Kai, mala and strengthening 140
NOTES TO CHAPTER 4 143
5 PARTICIPLES II: FORMULAS AND METRICS 1515.0 Introduction 1515.1 Onformulas 152
5.1.1 Formular language and ordinary language 1545.1.2 Formula and metre 158
5.1.3 Types of recurrence 1595.2 Formulas and form: the localization of formulas 164
5.2.1 The dactylic hexameter 1655.2.2 The localization of' participle + pef 171
5.3 Formulas and meaning: the integration of formulas 1865.3.1 Levels of integration 1875.3.2 Prototypicality and the semantic integration of formulas . . 189
NOTES TO CHAPTER 5 197
6 CONCESSIVE CONDITIONALS 2056.0 Introduction 2056.1 Concessives, conditionals and concessive conditionals 2066.2 Kinds of concessive conditional 2076.3 Concessive conditionals in Homer 209
6.3.1 Per in concessive conditionals 2096.3.2 Kai in concessive conditionals 2196.3.3 Ei per and kai ei in a diachronic perspective 224
6.4 Concessive and non-concessive conditionals 2276.4.1 Two kinds of neutralization 227
6.4.2 The two senses of ei per: 'even if and 'if really' 2296.4.3 From concessive to non-concessive eiper 230
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 8/318
Viii
CONTENTS NOTES TO CHAPTER 6 233
7 PERIPHERAL INSTANCES 2397.0 Introduction 2397.1 Linguistic irregularity in Homer 239
7.1.1 Unclassifiable vs. divergent instances 2397.1.2 The explanation of peripheral instances in Homer 240
7.2 Semantic integration of formulas again 2437.2.1 Formular or linguistic irregularity? 246
7.3 Divergence: changes in form and meaning 2497.3.1 Changes in meaning 2497.3.2 Changes in form 2527.3.3 From prototype to periphery 256
7.4 Per and inclusive focus particles 2577.5 At the outer periphery 2597.6 From Homeric to Attic per 262
NOTES TO CHAPTER 7 267
8 OVERVIEW 271
APPENDIX 277
REFERENCES 281
INDEX OF SUBJECTS 293
INDEX OFNAMES 305
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 9/318
PREFACE
This study was submitted as a doctoral thesis to the Faculty of Arts of theUniversity of Leiden. Originally, the description of the use of the particle perin Homer was undertaken as part of a larger project, the description of thediachronic development of per from Homeric to Attic Greek. After some
preliminary research, however, it appeared that, owing to the specific problemsthe descriptive linguist is faced with in Homer, there were reasons to give thedescription of the Homeric material a thematic unity of its own. Accordingly,the present study has to be viewed as a pubhcation in its own right, rather thanas a piece of subsidiary research.
Professor C.M.J. Sicking, the supervisor of the project, has followed the progress of my work from the beginning. Many of the adjustments which I havemade in the presentation of the argument are due to his often penetratingcriticism. I wish to thank Professor R.S.P. Beekes also for a number of usefulcomments and suggestions.
Professor C.J. Ruijgh read the whole manuscript in an early stage. Notonly did he save me from a number of factual errors; he also influenced the
spirit of the work, by making me attentive to the fact that a too rigid approachto the facts of language may impair the effectiveness of studies like the presentone.
I am indebted to the Foundation of Linguistic Research of the Netherlands Organization for the Advancement of Pure Research (Z. W.O) for a grantwhich enabled me to carry out all the preliminary investigations and to writethe first draft.
The Greek text in this book is presented in italics. The motivation for this practice, with which some readers may be unaccustomed, is the desire not toset up an unnecessary barrier for many readers who have primarily or exclusively a linguistic interest and whose judgment I value highly. The Greek word ac-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 10/318
χ PREFACE
cent was lost in the process, too, owing to a deficiency in the computer character set which was at my disposal. But for those who have swallowed the first inconvenience this will not be an unsurmountable problem either.
Leiderdorp, april 1988
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 11/318
1 INTRODUCTION
1.0 Aims of the investigation
The purpose of the present study is to provide a description of the Greek particle per as it occurs in the text of Homer. As such it is a contribution to thestudy of Ancient Greek in general, and of the 'Greek particles' in particular.But there are some circumstances on account of which the present work transgresses the boundaries of 'Greek linguistics' proper. Firstly, per appears to bewhat may be called a 'scalar particle', and the framework which I have devisedfor its description in terms of 'scalarity' cannot help contributing to the discussion of scalar phenomena in general in language.
Secondly, the study of per (as well as of many other linguistic items) in Homer cannot ignore the many problems and complications that are posed bythis author. 'Homer' is (i) a metrical text and has (ii) an oral-formulaic origin.These two properties determine to a high degree the linguistic form of the Iliadand the Odyssey. Thus, as a description of a linguistic feature in Homer, the
present study is also an essay in the relation between linguistics on the one handand formulas and metre on the other.
1.1 Per in Homer and after
The particle per is one of the salient linguistic features of Greek epic, i.e.Homeric, poetry. Its salience is partly due to the fact that the meaning and useof per in Homer is markedly different from the meaning of the particle in post-Homeric Greek. Per belongs to the linguistic items which mark the peculiarand deviant nature of Greek epic language.
An easily observable difference between per as it occurs in Homer and peras it occurs in later Greek is its distribution: in Homer the distribution of per
is very free; the particle can be attached, as an encUtic, to words belonging to
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 12/318
2 LINGUISTICS AND FORMULAS IN HOMER
very diverse categories: adjectives, participles, pronouns, substantives, and adverbs, as well as relatives and subordinating conjunctions. In later Greek, onthe other hand, the only productive expression-type in which per occurs is thefixed combination of the particle with relatives and subordinating conjunctions(subordinators), the most frequent combinations being 'relative + pef (hos
per) and 'conditional subordinator + per 9 (eiper). Now the statement of bare distributional facts is in itself of little avail as
long as it cannot be shown how the distribution of a linguistic item is related to
its meaning and its function on sentence-level and in discourse. So a mere distinction between a 'free' and a 'fixed' distribution of per is in itself insufficientto account for the difference between the Homeric and the post-Homeric useof the particle. There must be a difference in meaning as well. The first thingwe observe in the field of 'meaning' is that in Homer many expressions in which
per occurs have a concessive adverbial function in the sentence (see 1.3.2 below). In post-Homeric Greek, on the other hand, 'concession' applies onlyto a limited group of instances, and these instances may be seen as the continuation of Homeric usage.
It is possible, then, to divide, provisionally and intuitively, the total sumof instances of per in a binary way, into 'concessive' and 'non-concessive' instances. But this distinction only partially agrees with the distributional distinction mentioned above: all instances of post-Homeric 'subordinator + pef arenon-concessive, but the converse does not hold: not all instances of per inHomer are concessive. The asymmetry is due to the fact that the non-concessive use of 'subordinator + pef is not confined to post-Homeric Greek: it occurs in Homer as well. Thus, when we want to refer to 'concessive' per as'Homeric' per (as I do in the present study), we have to realize that this termdoes not apply to all the instances of the particle in the Iliad and Odyssey: out
of the total number of 534 instances of per in Homer, 385 are to be called'Homeric'; the rest are instances of the non-concessive 'subordinator + pef.In the present study only 'Homeric' per will be dealt with. The use of per
in subclauses (adhering to the subordinator: hos per, ei per) will be discussedonly in so far as it is relevant for the study of Homeric per. That is, of the instances of 'subordinator + pef only those will be discussed that yield a concessive subordinate clause (relative or conditional, see exx. (3) and (11) in 1.2
below). The discussion of the post-Homeric and non-concessive use of per insubclauses will find its proper place in a study on relative and conditionalclauses in Attic Greek which I am currently preparing (Bakker: in prep.).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 13/318
INTRODUCTION 3
The decision to treat Homeric and post-Homeric per ;separately may seemarbitrary, since it is made on the basis of a semantic distinction ('concessive' vs.'non-concessive') which was called above 'provisional and intuitive'. But whenwe assign 'concession' its proper place and describe it within a diachronic
perspective as the consequence of the proper meaning of per rather than thatmeaning itself, our bipartition in the material will appear to be better founded.In the course of the argument (see 3.2.3 and 4.2.1) I will show that 'concession'is a natural consequence of what will be referred to as 'scalarity' (see ch. 2).
And once we speak in terms of 'scalarity', the term 'Homeric per' turns out tocover a coherent set of linguistic phenomena.This is not to say, however, that there are no connections between
Homeric (scalar) and post-Homeric (non-scalar) per. It is possible in thesynchronic dimension to devise a semantic framework in which both scalar andnon-scalar per have their place and which, accordingly, reveals the commondenominator of both kinds. This framework will be presented in section 2.3.2
below.
1.2 Facts from the data-base
The particle is the central part of a number of more or less regular ex pression-types. In the present section I present the most important of these,each time giving a highly characteristic ('prototypical', see 1.4 below) example,to which a brief descriptional commentary has been added. The order of theexamples reflects the order in which the various uses will be discussed in chapters 3-7. The last two examples, (15) and (16), will not be discussed in the
present study; they are instances of post-Homeric, non-scalar per. Throughoutthis study the translations of the examples have been taken from Rieu (1946,1950). Whenever a more literal or explicit rendering than Rieu's was more ap
propriate, this has been added, in square brackets.
( 1) tô de Phobos philos huios hama krateros kai atarbês
hespeto, hos t'ephobêse talaphrona per polemistên."(Ares the Killer and his son), the fierce and indomitablePanic-maker, before whom the staunchest warrior turns tail". (Π .
13,300).
(2) smerdale' eurôenta, ta te stugeousi theoi per.
(The Netherworld): "The hateful Chambers of Decay that fill even
the gods with horror". (R 20,65).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 14/318
4 LINGUISTICS AND FORMULAS IN HOMER
In (2) per adheres as an encUtic to the substantive theoi ('gods') and in (1) tothe adjective talaphrôn ('brave'). Both collocations occur in a non-restrictive('digressive') relative clause whose relative subordinator is modified by the particle te. This particle imposes digeneric modality on the relative clause. In 3.11will show that to recognize this modality as such is important for the description of per as it is used in (1)-(2).
(3) kai d'Akhileus toutôi ge makhêi eni kudianeirêi
errig' antïbolêsai, ho per seo pollon ameinôn."Even Achilles fears to meet this man in the field of honour, who isa better man than you by far" [even though he is..] (77. 7,113-4).
Here it is per that modifies a non-restrictive relative clause, not te. Per adheresto the relative pronoun (ho(s)per). The relative clause has as its head noun the
proper name Achilles which is preceded by the particle kai ('even'). It is important to notice that the relative clause may be interpreted as concessive ('thoughhe is a far better man than you'). The expression-type exemplified by (3) will
be discussed in 3.2.
(4) speusomen ai ke nekun per Akhillêï propherômengumnon: atar ta ge teukhe' ekhei koruthaiolos Hektôr."We might at least save the body for Achilles, naked though it is.The armour at any rate is in the hands of Hector with the GlancingHelm". ( . 17,121).
(5) alla se per proetô."Let him at least allow you to take the field". (II. 11,796).
In these examples the addition of per to nekun ('corpse') and se ('you') resp.has approximately the same effect as at least in English. For a full understandingof the function of per in the expression-type exemplified by (4) and (5) we haveto notice that (4) and (5) are adhortations/commands. See further 3.5.
(6) all' oud' hôs hetarous errusato, hiemenos per."But he failed to save his comrades [even so he did not save...'], inspite of all his efforts". (Od. 1, 6).
Here per is attached to a participle. The combination 'participle + per issimilar to the relative clause in (3), in that both combinations are attached to
a constituent which is modified by even;3 in (3) this constituent is a propername; in (6) it is a demonstrative pronoun (hôs). But the similarity between (3)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 15/318
INTRODUCTION 5
and (6) is not limited to syntax. Semantically, too, they have a common element: just like the relative clause in (3), the participial phrase in (6) allows of a concessive interpretation. The expression-type will be further discussed in 4.2.
(7) alla ta men protetukhthai easomen, akhnumenoi per.
"However, what is done is better left alone, though we resent it still".
( . 18,112).
(8) hoi de kai akhnumenoi per ep ' autôi hêdu gelassan
"The rest, disgruntled though they were, had a hearty laugh at hisexpense" ( 2,270).
(9) Aineia, khalepon se kai iphthimon per eonta
pantôn anthrôpôn sbessai menos."Aeneas, powerful as you are, you can hardly expect to put everyonewho meets you out of action". (II. 16,620).
(10) Here, mê dê pantas emous epielpeo muthouseidêsein: khalepoi toi esont' alokhôi per eousêi
"Here, (the father of men and gods replied) do not expect to learnall my decisions. You would find the knowledge hard to bear,although you are my Consort". ( . 1,546).
These four examples differ from (6) in that the constituent which is modified by even (oud'hôs) is absent. They belong to the most common expression-type,which will be referred to as 'participle + pef. In this expression-type per adheres to a circumstantial participle. The combination has a concessive adver
bial function in the clause. The four examples may be differentiated in variousways. In (7) and (8) the participial phrase is a simple (middle) participle; in (9)
and (10), on the other hand, it consists of a nominal element (modified by per )and a participial form of be. The first kind of participial phrase ('participlealone') I call synthetic and the other ('nominal + participle of be') analytic.Within the analytic kind, moreover, a further differentiation is possible: thenominal modified by per may be either an adjective (as in (9): iphthimos,'strong'), or a substantive (as in (10): alokhos, 'wife'). For the distinction between adjective and substantive and the implications of this distinction for thestudy of per see 4.3 below. The last criterion on the basis of which (7)-(10) may
be differentiated is the presence or absence of the particle kai: in (8) and (9)the participial phrase opens with kai; in (7) and (10) it does not. The pattern in
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 16/318
6 LI NGUISTICS AND FORMULAS IN HOMER
the distribution of kai will be discussed in 4.5 and 5.3 below (see also 1.3.2 and1.5 in the present chapter).
( 11) steuto gar eukhomenos nikêsemen, ei per an autai Mousai aeidoien, kourai Dios aigiokhoio."(Thamyris) had boasted that he would win in a singing-match evenif the Muses themselves would be his competitors" (77. 2,597).
This is an instance of the typically Homeric use of per in conditional clauses.
The kind of ei per which means 'even if, exemplified by (11), does not occurafter Homer. Extensive discussion of the 'concessive conditional' will be
presented in chapter 6.
(12) oikade per sun nêusi neômetha."Let us sail home with our ships." (77. 2,236).
(13) ho de peisetai eis agathon per"He ( = Achilles) will Hsten to you to his own advantage". (II. 11,789).
(14) kreissôn eis emethen kaipherteros ouk oligon per / enkhei."You are a stronger man than I and not a little better with the spear."(77.19,217).
The meaning of per in these examples is far less identifiable than in the otherexamples. It might even be asked whether per here has any meaning at all otherthan a vague sense of 'emphasis'. The present investigation will show that it isfalse to treat examples like (12)-(14) as representative instances of Homeric
per. Chapter 7 deals with them.
(15) ta gar phroneeis ha t' ego per."For you have the opinion that I have". (77. 4, 361).
(16) Arêtê d'onom'estin epônumon, ek de tokêônton autôn hoi per tekan Alìdnoon basilêa."Her name is Arete and she comes from the same family as Alcinousthe King". (Od 7,54-5).
These are Homeric instances of 'post-Homeric' per (see 1.1). They will not bediscussed for their own sake in the present study (but see 2.3.2 below) and have
been included merely to complete the picture. Full discussion of per in 'thesame as-contexts' in Bakker (1986, in prep.).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 17/318
INTRODUCTON 7
1.3 Per in Greek linguistics
Per has been the subject of a number of discussions which can mostly befound in larger handbooks. I mention Chantraine (1953: 320-1), Denniston(21954:481 ff.), Kühner-Gerth (1904:168 ff.), Monro (21891:320-1) and Monteil (1963:160 ff.). Fraenkel's (1925:6-13) account is an article in its own right.These discussions can hardly be called successful descriptions of per in Homer.They are inadequately adjusted to the data-base in question, the range of phenomena which is covered by them being either too narrow or too wide. In
the present section I briefly discuss them. I shall assess them as to their abilityto answer the preliminary questions that should, in my opinion, be asked.
1.3.1 Emphasis
A number of proposals can be taken together under the general heading'emphasis'. Owing to its relatively free distribution, per has been said to'highlight' individual constituents in the sentence, or to emphasize their importance in a particular context. Possible candidates for an analysis in these termsare exx. (1)-(2), (4)-(5) and (12)-(14) above.
The general term 'emphasis' may be further specified. According toMomo ( 1891:320), for example,per is 'intensive', "denoting that the word towhich it is subjoined is true in a high degree." Closely related to this approachis the 'intensive' use of Denniston (1954:482), on which see more below (1.3.4).Denniston distinguishes more uses, of which the 'determinative' use is at themoment the most important. Denniston differentiates the determinative fromthe intensive use in the following way: "The particle denotes, not that something is increased in measure, but that the speaker concentrates on it to the exclusion of other things." Thus, we find two distinct 'values' of per in the
accounts which advocate the emphasis approach: intensification (of the meaning of a word), or highlighting (of the role/importance of a word in a particularcontext).
The trouble with this kind of approach is twofold. First, the notion of'emphasis' gives rise to considerable demarcation problems, particularly withregard to the particle ge, which, likewise, is said to have an emphasizing function. In other words, the range of the description in terms of 'emphasis' maywell comprise more examples than is desirable. Second, the concept of'emphasis' is in itself problematic, both from a definitory and from a linguistic
point of view. The following citation from Kühner-Gerth (1904:168) may serve
as an illustration of both problems:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 18/318
8 LINGUISTICS AND FORMULAS IN HOMER
(17) "Per stimmt mit ge darin überein, dass es, wie dieses, eine Mehrungausdrückt; weicht aber darin von ge ab, dass es die Mehrung extensiv (alsodas Mass, die Grösse, den Umfang des Begriffes), ge dagegen die Mehrungintensiv (also den Grad, die innere Stärke eines Begriffes, so stark er ist) bezeichnet. Per dient also eigentlich dazu, einen Begriff seinem Umfangenach hervorzuheben, sodann auch überhaupt einen Begriff hervorzuheben",
The gist of this unclear passage seems to be that an original meaning of per,in which per somehow reflects some property of the 'thing' ("Begriff') referredto, has evolved into an emphasizing ('highlighting') one. This distinction seemsto be similar to Denniston's distinction between an intensive and a determina-
Q
tive use of the particle. But whether we speak of the emphasis which is lent by per and ge to a given constituent 'X' in terms of a certain property of the referentof 'X', or in terms of the importance of 'X' in the sentence (which are, obviously, two very different things), we do not throw any light on the demarcation of
both particles.Kühner-Gerth's differentation of per and ge cannot be called successful,
as it is very hard to operationalize it. I doubt whether there exists a contrastive pair 'Xper and 'Xge' in which the addition of per n b unequivocally attributed to the size of X and the addition of ge to its 'degree'. Besides, a lot oflinguistic problems are involved in Kühner-Gerth's idea of "Mehrung". For example, are we to assume that the addition of per to a constituent is the expression of an objective property of the referent of 'X', or is it dependent on whatthe speaker wants to express? In the second case we essentially take thethoughts of the language user about what he says (or the way he envisages thethings he refers to) as the explanation of how he expresses his thoughts in language. But the only way to know these thoughts is to look at how they are ex
pressed. Clearly, there is a dangerous core of circular reasoning in this
approach.But in the case of the kind of emphasis which is connected with the im portance of a given constituent in a given sentence, it is not easy either to differentiate per from ge within the framework suppUed by the authors mentionedabove. According to Monro (21891:320), for example, the differentiation runsas follows: "Ge is used, like per , to emphasize a particular word or phrase. Itdoes not however intensify the meaning, or insist on the fact as true, but onlycalls attention to the word or fact, distinguishing it from others." What isdescribed here as the difference between per and ge is exactly the difference
between Denniston's intensive and determinative uses of per. Consequently, itmay be asked what the difference is betweenge and this determinative use (orMonteil's central meaning of per, see note 5). As long as the vague term
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 19/318
INTRODUCTION 9'emphasis' is not further specified than 'intensification' or 'highlighting', it isunlikely that a satisfactory answer to this question can be attained.
1.3.2 Concession and strengthening
The second major approach to Homeric per can be dealt with under theheading 'concession'. Concession is commonly considered to be the centralmeaning of per; this is understandable, for the great majority of the instancesof Homeric per occurs in circumstantial participial phrases which have a con
cessive adverbial function in the sentence, see exx. (6)-(10) in 1.2 above. Thecombination of per with a participle is the most frequent and conspicuous ex
pression-type of per in Homer. But 'concession' appUes to other, less frequent, but no less characteristic, uses of per as well, notably to the use of per in non-restrictive relative clauses (see ex. (3) above) and in conditionals (see ex. (11)).
The kind of problem connected with the term 'emphasis' does not occurin the case of 'concession': the application of the term is reasonably clear.'Concession' does not pose the kind of demarcation problems that are connected with 'emphasis' either: the instances to which it applies form a well-
defined group; there is no risk of including instances of ge. However, otherquestions arise. In itself the statement that 'participle + per' is a concessive adverbial element may be considered uncontroversial, but (i) what is exactly thecontribution of per to the concession, and (ii) how are the concessive instancesof per related to the substantial minority of seemingly non-concessive instances (for instance (1)-(2) and (4)-(5) above)?
In regard to the first question, the position commonly held may bedescribed as follows. Between a circumstantial participle and the clause towhich it is attached a number of semantic ('logical') relations may be distinguished, one of which is 'concession'. A concessive relation between a par
ticiple and a clause can be strengthened by particles, one of which is per. Thus,to say that per is concessive means essentially that per strengthens the concessive relation between a participle and a main clause.
This kind of description, in which the logico-semantic relation of'concession' is considered to be 'prior' to per and its function, is especiallycharacteristic of French literature: Chantraine (1953: 320-1) discusses per inconnection with concessive participles, implying that a concessive participlemay or may not be accompanied b y per. Likewise, Oguse (1962:133-143) discusses concessive participles in the first place, the presence or absence of per
apparently being immaterial for the concession. Monteil's (1963: 163) posi-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 20/318
10 LINGUISTICS AND FORMULAS IN HOMER
tion, finally, is the most explicit: "Une nuance logique est ainsi tout au plussouligné par per; elle n'est pas exprimé lexicalement par la particule."
However, this account of the concession in 'participle + per 9 is contestable, not only from the point of view of per and its description (as we shall see),
but also from the point of view of the description of the circumstantial participle in Ancient Greek. To posit that a particular logico-semantic relation between a participle and a main clause is expressed by the participle isquestionable, since participles, unlike adverbial subclauses, simply have no
means to express a given relation lexically and overtly. The crucial feature of participles is precisely that they are unspecified as to their relation with themain clause (as well as their dependence on the main clause), see further 4.1.
Thus the function of per in partiicipial phrases has to be reexamined. If the participle does not by itself express the concessive relation, the description of per in the participial phrase as a strengthener loses its point too. An investigation of the division of labor within the participial phrase involves, furthermore,an account of a second particle: kai (see exx. (8)-(9) in 1.2 above). When kai is
present, it usually precedes the participle, while per follows it. Some authorstreat kai in these situations as a more or less meaningless strengthener (thestrengthened element being per), while others attribute to per and kai distinctmeanings/functions. The demarcation problems which are inherent in thissituation require a systematic description of the respective roles of per and kaiin the participial phrase.
The instances of per to which 'concession' applies are the most frequentin the data-base, but there is a substantial minority of instances of Homeric perthat do not occur in a concessive adverbial phrase (participle or subclause).This leads to the second question mentioned above: how are the concessive in
stances related to the other instances? If this question cannot be satisfactorilyanswered, the approach to per in terms of 'concession' will turn out to yield adescription with a too narrow scope: it applies only to a part of the material,leaving in the dark what the instances belonging to this part have in commonwith the remaining instances.
But also when one considers concession as a separate 'sense' of per (seeDenniston 1954:484), rather than as its central meaning, one has to state whatthis 'sense' has in common with the other senses, for the very point of a divisionin 'senses' is that there is one property which all the senses have in common.This common property, however, is not provided by Denniston.16
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 21/318
INTRODUCTION 11
In fact, the two questions considered in this section are interrelated. Alack of precision in the determination of what per contributes to the concessionmay easily lead to an account of per in which the relation between the concessive and the non-concessive instances is neglected. In general, we may say thatmost accounts of per fail to bring unity and coherence anyway into the description of the wide array of uses of Homeric per as presented in 1.2. In (not) doingthis, they do no justice to what seems to me, generally, an importantmethodological principle in descriptive linguistics, viz. unification, the assump
tion that linguistic items, as long as the contrary has not been proved, have one basic meaning, which is the common denominator of all the different senses oruses of the item. In the course of this study, I shall argue (and show) that a systematic description of per has the unifying effect which is lacking in the accountsmentioned.
133 Fraenkel: "per heisst 'auch"'
One account of per deserves to be discussed in its own right, rather thanunder a general heading. This is Fraenkel (1925). Despite some dubious as
sumptions and questionable steps in the argument, Fraenkel's account is theonly one in which per is discussed, albeit implicitly, as what it actually is, viz. a focus particle (see 2.3). Fraenkel straightforwardly equates per with German'auch' ('also'): " per heisst 'auch'". This equation leads him to assert that theconcession is a consequence of the presence of per, rather than a separate senseoftheparticle(1925:7):
(18) "Der Inhalt des Partizips steht im Gegensatz zu dem übrigen Satze, und das per besagt ausdrücklich, dass 'auch' unter diesen Umständen die Sache so istoder geschah. So bekommt das ganze Gefüge einen Konzessiven Sinn, undda es zum habituellen Schema wurde, konnte diese Bedeutung sich fest darineinwurzeln. An ihr erhielt nun auch das Wort per seinen Anteil, und erwarbso die Fähigkeit, konzessive Farbe leicht anzunehmen, obwohl sie ihm vonHaus aus nicht eignet."
The advantages of this approach are considerable. Not only is the role of per inthe expression-type participle + per' better described this way (both from asynchronic and from a diachronic point of view); the description of the concession as the standardized consequence of auch has also an immediate unifyingeffect. A number of instances for which the other descriptions have no bettersolution than the emphasis approach (see 1.3.1), such as (1) and (2) in 1.2, are
better accounted for when we conceive of per as 'auch'.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 22/318
12 LINGUISTICS AND FORMULAS IN HOMER
However, Fraenkel's approach raises considerable problems as well. Forone thing, the apodictic equation of per with auch may suggest that both particles have the same semantics and distributional properties. But this is not so.A given instance of 'auch X' is not without more ado equivalent to 'Xper' Theequivalence is conditional; it depends on a number of factors. And only whenwe have determined these factors are we on our way towards a real description.When viewed this way, Fraenkel's account is not a description at all, but a 'pre-theoretical' equation of per with a word which simply happens to fit many pas
sages well in a translation.The second problem one encounters when dealing with Fraenkel's account is connected with the first one; it concerns, again, the delimitation oiperwith regard to the particle kai. When per is 'auch', what are we to do with kai,which is, after all, the normal Greek expression of 'auch'? That both particleshave the same value is for Fraenkel beyond dispute. Rather, he contends thatit is the frequent collocation of both particles that should be accounted for(1925:12):
( 19) "iWie kann per, wenn, es 'auch' bedeutet, so oft neben dem gleichbedeutenden
kai stehn? Nun, doppelter Ausdruck desselben Begriffes oder derselbenVorstellung ist stets im Sprachleben üblich. Die Wörter haben wie dieMünzen nicht nur ein Stempel und Sinn, sondern auch ein Gewicht; undwenn sie abgegriffen sind, werden sie leichter. So muss man, wenn man vollzahlen oder reichlich zahlen will, oft noch zulegen." (Emphasis added).
On the basis of the assumption that per and kai have the same meaning,Fraenkel asserts that their frequent collocation has to be accounted for in termsof strengthening. Admittedly, the notion of 'strengthening' covers a set of
phenomena in actual language use which cannot be ignored. And it applies tomany collocations oiper and kai, for example, (8) and (9) in 1.2.18 But a closer
look at the data-base reveals that the relations between per and kai are muchmore subtle and diverse than Fraenkel claims. There are many cases in which
per and kai have functions that clearly differ from one another, for instance exx.(3) and (6) above. To recognize these cases as such and distinguish them fromthe cases where mere strengthening obtains and/or where per and kai are interchangeable is one of the main tasks with which the investigation of per isfaced.
1.3.4 Etymology and diachrony
Among all the uncertainty and disagreement about the meaning of per,there is one uncontroversial point: the particle has a clear etymology. But this
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 23/318
INTRODUCTION 13
raises new problems, for the meaning associated with this etymology cannoteasily be exemplified, according to some authors (Fraenkel, Denniston), on the
basis of the instances of per as we find them in Homer. In itself a disconnection between the actual use and the original (etymological) meaning of a given linguistic sign need not be a problem for the descriptive linguist.19 But the pointin the present case is that there is no apparent disconnection.
In itself the etymology oiper may be considered perspicuous. It is generally accepted that per is cognate to perì. The preposition peri normally means'all round'. It is often used adverbially in Homer (see Monro 21891:172) withthe meaning 'exceedingly'.21 As a preverb peri most often means 'around', butit may have the 'intensive' sense as well: peri-einai, peri-gignomai - 'be superior', 'surpass'). The 'intensive' meaning may be attributed toper- as a prefixin Latin: permagnus, perclarus ('very big, clear'). Accordingly, the originalmeaning of the particle per is supposed to be in the superlative sphere, too. Theinstances of per which allegedly best reflect this 'original' use are grouped byDenniston (1954: 481) together under the heading 'intensive use' (on whichsee also 1.3.1).
Denniston's intensive examples (one of which is (14) above) form a very
limited group which stands more or less isolated among the other uses he distinguishes. He therefore posits a diachronic gap between the original,etymological use on the one hand and the extant, observable uses on the other,thereby implicitly treating his 'intensive' instances as relicts of the earlier, extinct use: "The use of the particle which best supports this etymology, its intensifying use with adjectives and adverbs, is in fact the most difficult of all toexemplify. The primary use must have been metamorphosed before the timeof the earliest extant Greek literature" (p. 481).
Similarly, Fraenkel disconnects the extant Homeric uses from the commonly accepted etymology: "die Zusammenstellung mit perì 'sehr' versagt ja
hoffnungslos gegenüber den wirklichen Gebrauchsweisen von per" (p. 12). Evidently, an intensifying ("urgierend") use oiper is compatible with Fraenkel'sequation oiper with auch. He therefore dismisses it altogether, positing insteadthat the meaning 'auch' is semantically derived from 'darüber hinaus'('beyond') as a separate sense of peri. This solution is ingenious, but gratuite inso far as it depends on 'auch' as the meaning oiper.
Now we may say, as a general principle, that claims about the diachronyof a given item have to be based on synchronic research. Diachronic claims
presuppose synchronic claims, not vice versa. When we cannot state what two
items have in common in the synchronic dimension, we cannot do so either in
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 24/318
14 LINGUISTICS AND FORMULAS IN HOMER
the diachronic dimension. Accordingly, when we say that Denniston's andFraenkel's claim about the etymology and diachronic development of per isdubious, this amounts to saying that they have not described the extant uses ofthe particle in a satisfactory way. In the present study I will show that it is possible to reconcile Fraenkel's thesis (viz. that the meaning of per has to bedescribed in connection with 'focus particles' like 'auch') with the original intensifying meaning which is suggested by etymology. In other words, the centralsynchronic meaning of per in Homer combines elements from the kind of par
ticle to which 'auch' belongs and from 'superlativeness' and 'intensification'.The description of the expression-type to which exx. (1) and (2) in 1.2 belongwill occupy a central place in this reconciliation.
As for Denniston's 'intensive' instances, given that the majority of the instances of per can be described in such a way that no conflict with the apparentetymology arises, this group loses its justification as a separate category. Actually, some of Denniston's intensive cases can be easily accounted for withinthe framework presented in the present study, while others (like (14) above)are highly atypical. In ch. 7 below I will argue that they are not 'old' at all, letalone relicts of an extinct meaning.
Thus there is no need to assume a lost original meaning, because the examples by which it is allegedly exemplified can be accounted for otherwise. Butthis does not mean that the instances of Homeric per are, as regards diachrony,a homogeneous mass. Actually, there are clear diachronic relations between,and developments within, the various uses of per, 'Concession', for example,cannot be properly discussed outside the diachronic dimension, because it is inthe first place the result of something (viz. scalarity). That the authors dealingwith per (with the notable exception of Fraenkel, see (18) above) have not beenable to recognize this diachronic development is due, again, to their unsatis
factory accounts in the synchronic dimension.
1.4 Categories and prototypes
In the present section I introduce a principle which is indispensable in linguistic investigations like the present one, viz. the notion of prototypicality. Thisnotion involves a distinction between the central, characteristic (prototypical)instances of a given linguistic item and the less characteristic, peripheral ones.When speaking in terms of prototypicality, we conceive of 'having a certainmeaning or function' as a matter of 'more or less' rather than as a matter of 'yes
or no'. Prototypicality (or 'Prototype Theory') originates from cognitive psychology, where the insight has been developed that human categorization
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 25/318
INTRODUCTION15
(i.e. the grouping together, as a matter of cognitive economy, of a range ofseparate phenomena under one label) does not yield discrete, but continuousand overlapping categories (see Rosch 1973,1978).
The introduction of 'prototypicality' in linguistic studies may have a purely practical motivation. When we study the use of a given item in a given cor
pus of text(s) and devise a framework within which the instances of that itemcan be described or classified, we will often, if not always, discover that thereare instances to which the description does not, or not fully, apply, or which
defy clear-cut classification. These instances may be seen as a 'residue', whichis left after all the clear and easily classifiable instances have been dealt with.There are three things we can do when we have to deal with such a residue.
First, we can do no less than reject our entire descriptional apparatus. But thisis, all other things being equal, a very drastic thing to do, for the apparatus wasdevised not for nothing: it does after all fully apply to the majority of the instances in question. As a second possibility we might try to explain the residueaway or discard it as irrelevant. But this is not attractive either, for the veryreason that we have chosen to study the actual use ( = all the instances, not aselection) of the item in the corpus. The third solution lies in taking the existence of the residue for granted, thereby accepting what Givón (1984:12) ex
presses as follows:
(20) "Categories conform to their basic definition in the majority of cases, andrules obey their strict description more likely than not. But there is always acertain amount of messy residue left, one that does not seem to fit into thecategory/rule in the strictest sense of their definition."
The third solution is the right one when one is prepared to take the data-baseas seriously as possible. It involves positing a continuimi ('scale') along which
the instances of the item under study can be ordered by degree of decreasing prototypicality.Prototypicality in linguistic studies involves, I think, two concepts,
gradience and neutralization. These two concepts are interrelated, but theyshould be kept distinct. 'Gradience' applies to the internal structure ofcategories and to the semantic fields of linguistic items, while 'neutralization'may be used when we view the relations between linguistic categories in thelight of Prototype Theory. I start with 'gradience'.
In the traditional, logic-based theories of categorization, membership ofa category is a matter of 'yes or no'. The properties which determine inclusion
in a given category are stated in terms of a 'checklist', to the effect that, givenfour properties a, b, c and d, and a category X, any candidate member of 'X'
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 26/318
16 LINGUISTICS AND FORMULAS IN HOMERhas to possess all four properties if it is to be called a member of 'X'.27 InPrototype Theory, on the other hand, there is no place for checklists. Instead,membership of a category has to be stated here in terms of intersections: givenfour properties, a, b, c and d, and a category X, any member of 'X' may haveany one of the four properties in isolation, two of them, three or all four. Thisaccount of the internal constituency of categories automatically leads togradience, because the members of a category can be ordered by the numberof prototypical properties which they possess. Schematically (borrowed from
Givón 1984: 14):(21)
The dark area is the focal area, where the prototypical instances of the categoryhave to be located. These instances have all the prototypical properties. Theinstances that have only one property, by contrast, are peripheral;28 they formthe residue meant in (20) above.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 27/318
INTRODUCTION 17
To speak in terms of Prototype Theory does not only affect our accountof the internal structure of categories. It also has consequences for how we conceive of the relation between two categories. Since categories are known bytheir prototypical instances, we may say that to distinguish category 'A' fromcategory 'B' is to distinguish the prototypical instances of 'A' from those of 'B'.This is unproblematic. But how are we to conceive of the relation between the
peripheral instances of both categories? In the original field of appHcation ofPrototype Theory, cognitive psychology, it is customary to speak in terms of
overlap here: two categories may display a certain amount of overlap, to the effect that they are not only internally continuous, but also externally, with regardto each other. For this conception there is prima facie evidence, in the existence of, for example, peripheral instances of the category 'chair' which may becharacterized as peripheral instances of the category 'table' as well. In otherwords, in such cases there might arise controversy as to which category a givenobject belongs.
In the case of linguistic categories, on the other hand, matters are different. Members belonging to linguistic categories can be ordered by degreeof decreasing prototypicality in the same way as members of 'perceived worldcategories'. But the important thing is that this gradience cannot be perceived:as to their form all instances of linguistic categories are similar. In the categories'Verb' and 'Noun', for example, all instances are morphosyntactically markedas verbs and nouns, no matter how peripheral they are. Likewise, in lexicalsemantics, the most peripheral instance of a given lexeme is as to its form identical to the prototypical instances. It follows that 'overlap' as the characterization of the relation between two categories has much less application inlinguistics than in cognitive psychology. For in linguistics the peripheral instances are just as different from one another as the prototypical ones. There is no
question of an instance of a linguistic category being indistinguishable from instances of other categories.
Instead of 'overlap', neutralization seems to be the appropriate concept todeal with the border-line area between two linguistic categories. Two
peripheral instances of two neighboring categories are never indistinguishable, but we may say that in these instances the difference that exists between the prototypical cases of their respective categories is neutralized, or in any case
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 28/318
18 LINGUISTICS AND FORMULAS IN HOMER
minimized. Thus, peripheral instances of neighboring categories may be substituted for one another without essential differences in meaning, at least to afar higher degree than their prototypical counterparts. This is reflected inspeakers' varying opinions as to whether 'A' or 'B' is the right expression in agiven context.
It is not difficult to apply the prototypicality framework to the study of peralready at the present stage in the exposition of the argument. In 1.31 have argued that per has not satisfactorily been described. In the Ught of Prototype
Theory, this means that the prototypical properties of per have not been recognized as such. It is easy to see that this has serious consequences. When we areunable to distinguish the prototypical instances from the peripheral ones, ourdescription lacks conviction and runs the risk of losing all sense of direction.The demarcation problems connected with the 'emphasis'-approach which Ihave mentioned (see 1.3.1) are highly typical in this connection. For demarcation problems typically arise when we mistake the peripheral instances for the
prototypical ones.Thus it is not the existence of demarcation problems as such that should
be objected to, for the existence of the borderline cases that cause these problems is undeniable, see exx. (12)-(14) above, as well as the examples in ch.7. The point is that the demarcation problems need not, and should not, arisein the delimitation of per as a whole. For this delimitation is made on the basisof the prototypical instances, which are maximally distinct from the prototypical instances of the neighboring ge and kai. The business of describing themeaning of per amounts thus to describing the prototypical instances of the particle. In the process of this description, which will occupy the greater part ofthis study (chapters 2-6), a number of essential properties will emerge, whichwill serve as a set of prototypical features by which the degree of peripherality
of the remaining cases (the 'residue') may be determined (see chapter 7). Examples like (12)-(14) will thus appear to be a great deal less prototypical thanexamples like (1)-(2).32
1.5 Linguistics and formulas in Homer
Prototype Theory predicts that no description of a given linguistic itemcan ever be expected to cover all the instances of the item (in the case of a text-corpus as data-base) or all possible instances (in the case of 'living' linguisticmaterial), simply because not all members of a category are prototypical. This
applies toper too. But in the case of per there are some complications, becausethe nature of the data-base ('Homer') is such that it is likely to have effects
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 29/318
INTRODUCTION 19
upon the nature and composition of the residue, the group of less than prototypical instances. The complicating nature of 'Homer' as a data-base forlinguistic research is the subject of the present section.
The Iliad and Odyssey are oral-formulaic texts. Despite some complicating factors, such as the degree of sophistication in plot and characterization dis
played by the poems, this statement may be considered fairly uncontroversial(at least in the Anglo-Saxon world). The formularity of the Homeric poems is,equally uncontroversially, the consequence of the fact that they are the product
and very probably the culmination of a tradition of oral poetry. The oral provenance of Homer has important consequences for the linguistic form ofthe Iliad and the Odyssey, not only on the level of the phrase (word group), thelevel where most of the acknowledged 'formulas' have to be situated, but alsoon lower levels (the word) and higher levels (the verse, sentence, paragraph,theme, etc.). In the present study I am, as regards oral poetry phenomena,
primarily interested in the level of the phrase (formula) and of the verse.On the phrase/formula-level the implications, for linguistic studies, of the
oral-formulaic provenance of Homer may be stated as follows. A given phrase,which has a function as a formula in the diction, may at times be used by a poet,under formular pressure in oral performance, under semantic and syntactic circumstances for which it was not devised originally. In those cases, the less thanoptimal well-formedness of the resulting expression has to be discussed, strictly speaking, in terms of the use of formulas in epic diction, rather than in termsof ordinary syntax and semantics. I refer to the use of a formula in a context forwhich it was not originally devised as the 'imperfect semantic integration' ofthat formula (see further 5.2, 6.3.1.4 and 7.1).
Imperfect semantic integration is a useful concept in the linguistic studyof (Greek) epic poetry, but it must be added at this stage already that it is very
difficult to determine its exact role. Any alleged case of bad semantic integration of a formula may in principle always be a peripheral, less than prototypical instance of the phrase in question. And it is often impossible, to determinewhether 'irregularity' in epic poetry is due to the 'jeu de formules' in the oral performance or to the flexibility which is inherent anyway in language. It seemsto be a good policy, then, to treat '(bad) semantic integration' as a factor in thediscussion of 'irregular' instances, which may operate simultaneously with
prototypicality. It is dangerous to treat it as a separate explanatory label. Besides, there is the more reason to do so because the concept of 'formula' is initself a definitely gradient phenomenon. 'Being a formula' and 'not being a
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 30/318
20 LINGUISTICS AND FORMULAS IN HOMER
formula' are not mutually exclusive affairs, but the two extremes of an extensive scale.
On the more formal level of metrics and the verse, the basically oral nature of Homeric poetry is an important factor too. The dactylic hexameter isnot an easy verse-type for the Greek language. And the oral poet, in addition, is under constant time pressure: in oral composition the metrical spacehas to be rapidly filled without interruption. This may give rise to the existenceof 'stop-gaps' of various kinds, words or phrases whose use or length is metri
cally motivated. In 5.3.21 will show that the particle kai may be used as such aword. The distribution of kai in participial phrases with per (see exx. (8)-(9))can be explained within the framework of the set of rules and tendencies thatconstitute the dactylic hexameter.
But it does by no means follow that we are entitled to say that a given wordwhose occurrence is metrically motivated would not have been used outsidemetrical space. Metrically motivated words are often redundant: they do notcontribute anything essential to the (meaning of the) context in which theyoccur. But then words in ordinary, non-formulaic language are often redundant, too. Redundance is a typical property of instances belonging to the
peripheral area of semantic fields, and it is the common denominator of manystrengthening phenomena in language. Thus, it seems to be, again, a good
policy to avoid treating a given peripheral and 'redundant' instance {of per orkai or any other particle or word) as exclusively due to the dactylic hexameterand its exigencies.
Beside formulas and metrics there is a third important factor in the linguistic study of Homer. Homeric poetry tends to display the co-existence of oldand young elements to a higher extent than non-formular language. The typically epic projection of the diachronic dimension onto synchronic space has in
the last resort an oral-formulaic motivation: an element or expression-type belonging to older diachronic strata of the language may be preserved in thediction (especially when it is an old formula) because a contemporaneousequivalent (metrically and/or semantically) expression which can replace it hasnot yet been found. But 'archaism' may also become an end in itself, originating from a desire of poets to lend dignity to their diction.
The diachronic aspect of the Homeric language is an important subject inthe present study, because it is unlikely that the wide array of uses of per has
been synchronically productive in the ordinary language at one and the sametime. The knowledge that 'diachrony' is all-pervasive in the Homeric language
may be seen as a useful and even necessary background for the description of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 31/318
INTRODUCTION 21 per, in that it is conducive to the assumption that there are diachronic relations between, and developments within, the various uses. But 'epic diachrony'remains a mere foil for this assumption; I will not discuss it for its own sake,leaving aside the many intricacies which it possesses.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 32/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 33/318
NOTES TO CHAPTER 1
1) The 'concessive' instances of per in post-Homeric Greek are atypical in that they arenot instances of 'subordinator + per' ; they have the form 'kaţper + participle' (functioning as aconcessive adverbial element: even though).
2) See for this and other numerical facts the Appendix at the end of this book.
3) Oude in (6) is equivalent to kai in a negative context: not even. This is more accuratethan saying that oude is the negative counterpart of kai, see 2.4.1 below.
4) It is fair to note that the authors cited in this part of the doxography (with the exception of Monteil, see the following note) do not commit themselves entirely to the 'emphasis' approach: they distinguish other uses of per as well, notably the concessive one, see below, 1.3.2).
5) See also Monteil (1963:163-4): "Per est un moyen à la fois lexical et économique d'attirer
sur un mot l'attention, d'en faire un point d'orgue. (...) Ainsi compris, per est une sorte de signe
diacritique oral; il n'intensifie pas l'idée d'une qualité ou d'un procès, mais signale avec insis
tence l'importance, dans l'enchaînement des faits, d'une notion, quelle qu'elle soit." And Ruijgh
(1971:444): "Homère emploie parfois la particule (...) pour souligner un mot (ou un groupe de
mots) dans le cadre de la phrase."
6) The first value ('intensification') is the one which is associated with the etymology of
per. See 1.3.4 below.
7) Notice that Kühner-Gerth in this passage explicitly deny per the characterization
'intensive', attributing it instead to ge. This is very strange, for just before the passage cited they
say: "Die Grundbedeutung vonper ist: durch und durch, durchaus, aber nicht in Beziehung auf
den Raum, sondern auf den Grad {durchaus, sehr) und die Umstände {jedenfalls)" To my mind
this passage flatly contradicts the one cited in the text, where "Raum" ("Grosse") is attributed
to per and "Grad" (and, consequently, an 'intensive' meaning) to ge.
8) But Denniston does not posit a diachronic relationship between the two uses.
9) 'Participle + ge' is not a well-defined and fixed expression-type. And incidental instan
ces of 'participle + ge' do not by themselves have a concessive meaning.
10) See Goodwin (1889: 336), Kühner-Gerth (1904: 77).
11) See also Oguse (1968).
12) The idea that participles express the various semantic relationships is deeply ingrained
in many accounts of the Greek participle. Goodwin (1889:336) goes even as far as to call con
cession a separate sense of the participle. For 'sense' and the implications of the use of this term,
see below.
13) E.g. Kühner-Gerth 1904:170 and Fraenkel (1925:13).
14) Denniston (21954: 486): "Often the participial clause opens with kai. Kai.per then
means 'even though', kai marking the climax and per the concession", and Chantraine (1953:
320-1), more explicitly: "Les deux particules sont disjointes: kai marque la progression et per
l'opposition." Notice, incidentally, that distinct functions need not necessarily preclude
strengthening: 'even though' in English is 'stronger' than 'though', even though 'even' and
'though' have a different meaning.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 34/318
24 LINGUISTICS AND FORMULAS IN HOMER15) Of course, it is in principle always possible that two alleged senses simply have noth
ing in common. In that case we have to speak in terms of homonymy. However, the decision to
call two words homonymous (i.e. lexically entirely distinct) has to be postponed as long as pos
sible. As long as there is the slightest similarity, we should speak in terms of polysemy, the pos
session by one lexeme of two or more related senses. In practice it is often hard to determine
whether two identical words are homonyms or two related sense of one lexeme. Often diachronic
factors are involved. See Ruijgh (1971:102), Lyons (1977: 550 ff.) and in general Cruse (1986:
68 ff.).
16) Kühner-Gerth implicitly provide a common property, in that they assert that conces-
sive per emphasizes oppositions ("Zuweilen dient es ( = per) dazu, Gegensätze hervorzuheben").
But this solution suffers heavily, of course, from the opacity of the term 'Hervorhebung', see
above, 1.3.1.
17) My principle of 'unification' has much in common with one of the two types of diagram
matic iconicity in language distinguished by Haiman (1983:115-6), viz. isomorphism, the ten
dency for there to be a one-to-one correspondence between form and meaning of the linguistic
sign. Haiman's isomorphism is a reformulation of the classical Saussurean account of the lin
guistic sign. For a discussion of Haiman's isomorphism in the light of a modern branch of Saus
surean linguistics (viz. the 'Columbia School', or 'Form-Content analysis', see the articles in
Klein-Andrieu, ed. (1983)) see Kirsner (1985).
18) But mere 'strengthening' is not sufficient to fully account for the distribution of kai in
the expression-type 'participle + pef in Homer. See 5.3.2 below.
19) The etymological connections of a given item need not have any bearing on the actual
use of that item, and in the case of particles etymology can often be dismissed. This is why the
discussions of particles in Schwyzer-Debrunner (1950: for per. 553 ff.) is of no much use. These
authors are more interested in the etymology and Indo-European connections of the Greek par
ticles than in their actual use. Similarly, the subdivisions of the class of particles in Brugmann-
Thumb (1913: 607 ff.) are entirely etymologically motivated.
20) See Frisk (1960-72: s.v.), Chantraine (1968 ff: s.v.), Brugman-Thumb (1913: 629) and
many other authors. In Aeolic as well as in other dialects perì appears as per. This form is com
monly described as a derivation of peri (by way of apocope, see Buck (1955: 81)), but it may in
principle be a genuine old element.
21) It is not difficult to state what these two 'senses' have in common: "Die Vorstellungdass etwas rings umher geschieht und keine Richtung ausgeschlossen ist, ergab den Begriff der
Vollstandigkeit oder des hohen Grades." (Brugmann-Thumb 1913: 513).
22) Momo (1891:172) citi s boulêi periidmenai allôn ('knowing in council beyond others').
23) Cf. lso parumper,paulisper ('a (very?) short while'). For a discussion of the possible
connections between these forms and Greek minuntha per, see 7.1.1 below.
24) Exactly the same applies to Monro's (1891:320) collection of intensive instances. Some
of these are normal cases, and others highly uncharacteristic (i.e. lacking most of the important
properties of per). Denniston's and Monro's collection of intensive instances are, incidentally,
by no means identical. This fact alone should raise doubts as to the reality of this category.
25) One might ask how uses that have to be assigned to different diachronic strata come
to be used side by side in one and the same text. This question will be briefly discussed in 1.5below.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 35/318
NOTES TO CHAPTER 1 25
26) The typical escape from this problem in Transformational Grammar is the com
petence-performance- distinction. In this theory the 'fuzzy' and unclassifiable linguisticc
phenomena exist only on the level of 'performance' and are as such of no importance, since on
this level language is merely real and therefore imperfect. What really matters in this concep
tion is 'competence', viz. language seen as a set of exceptionless rules in the brain of the ideal
speaker. Criticism of the competence-performance distinction (and related theorems) in Givón
(1979: ch. 1) and in Baker & Hacker (1984: 281 ff.). In Givón (1982) it is shown that logically
based explanations of and approaches to natural language facts tend to apply only to the most
extreme (prototypical) cases. But the phenomena that are not amenable to logical analysis are
very numerous in any language and cannot be ignored.
27) This checklist constitutes, in logical terms, a set of necessary and sufficient conditions.
For a discussion of 'checklist' theories of meaning and prototypicality see Fillmore (1975).
28) The four properties that determine the category may, of course, be hierarchically or
dered, to the effect that a member possessing property 'a' may have a higher degree of
prototypicality than a member possessing 'b'.
29) Givón (1984: 14-7) describes prototypicality as a compromise between two extreme
theories of categorization. On the one hand there is what he calls the 'Platonic' point of view, in
which all the members of a category have all the criterial properties of that category. The other
extreme is the 'Wittgensteinian' point of view (characterizable in terms of 'family resemblance'),
in which often none of the members possess all properties, as in the case of a category with the
following members: ABC, BCD, CDE, DEF, etc. In Prototype Theory some of the members
have all the properties, the others being peripheral.
30) For prototypicality and the categories 'Verb' and 'Noun' see Hopper & Thompson
(1985).
31) Sometimes (e.g. König 1986) the concepts of 'overlap' and 'neutralization' are used
indiscriminately. In the light of the above argument, this usage turns out to be imprecise.
32) Notice that prototypicality does not only apply to the semantic field of per as a whole,
but also to specific expression-types in which the particle occurs. It will be possible, for example,
to define the prototypical features of 'participle + pef and to set off the prototypical instances
of that expression-type against the less prototypical ones (see 4.3).
33) The work of Milman Parry was pioneering here: Parry (1928a) explores and proves
the existence of formulas in Homer, and Parry (1930,1932) explains the existence of formulasby ascribing them to the oral nature of 'Homer'.
34) Note that this idea is in itself by no means a new one: from the preface of Chantraine
(1953): "L' Iliade et I'Odyssée sont (..) des oeuvres littéraires qui se sont développés par les
procédés d'un style oral, au moyen de la technique des formules épiques. H en résulte que telle
formule, employée en un passage donné, peut être transférée ailleurs où elle convient moins
bien."
35) This is one of the reasons why there is still no definitory agreement on the concept of
'formula'. See 5.1 below.
36) Throughout this study my approach to hexameter metrics will be synchronic and not
diachronic. Consequently, I will refrain from the vexed question concerning the origin and the
Indo-European connections of this verse-type.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 36/318
26 LINGmSTICS AND FORMULAS IN HOMER37) We may say that, as a general rule, an element whose function it is to strengthen another
element, like kai in participial phrases, is a peripheral instance of its category.
38) The diachronic aspect of Greek epic diction is a very complicated affair, which is in
timately connected with the various dialectal stages (Mycenaean, Aeolic, Ionic) through which
the epic tradition has gone, 'Diachrony' affects all the domains of the language system: syntax
and semantics, morphology and phonology. For a seminal discussion see Parry (1932). See also
Chantraine (1948: 89 ff.) and Ruijgh (1957). In Bakker (1988b) I discuss some phonological
phenomena in the light of epic diachrony.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 37/318
2 THE LINGUISTIC BACKGROUND: SCALARITY
2.0 Introduction
In the present chapter I present the theoretical background for the discussion of per as a scalar particle. The notion of scalarity appears to be the keysemantic concept in the study of Homeric per. To recognize it has a strong unifying effect on the description; it provides the link between the numerous concessive instances of the particle and a number of other instances which areunrelated at first blush (see for 'unification' 1.3.2 above). 'Scalarity' is the singlemost important prototypical feature (see 1.4) of per, instances of per that arenot scalar are definitely more peripheral than instances that lack other, less im
portant, prototypical features. The scalar backgroimd of the description of perinvolves, among other things, an account of the particle even in English. Thesemantics of even are to some extent cognate to that of per, and a number ofexpression-types to which per belongs cannot be properly discussed without
reference to even and its semantics, simply because the particle kai, the Greekequivalent of even, plays a role in them.Furthermore, an account of the semantics of the scalar particle even is
necessary for the delimitation of per and kai. As a preliminary to this delimitation, I introduce in the present chapter (2.2.2) the distinction between'intension' and 'extension'. As will be shown, this distinction lies at the basis ofthe difference between even and the so-called 'scalar superlative' (see 2.1.2).This difference, in its turn, points the way to the delimitation of per and kai inchapters 3 and 4.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 38/318
28 LINGUISTICS AND FORMULAS IN HOMER2.1 Scales
2.1.1 Scalarity: introductory
Consider, as a start, the following example ( = ex. (2) in 1.2 above):
(1) smerdale' eurôenta, ta te stugeousi theoi per.(The Netherworld): "The hateful Chambers of Decay that fill the(very) gods with horror." (iZ. 20, 65).
On first sight we have here a simple statement about the Netherworld: it is asserted, in a non-restrictive ('digressive') relative clause, that Hades is feared by the gods, just as we might say about Mt. Olympus that it is inhabited by thegods. However, the statement of this simple fact is clearly not the intendedmeaning of (1). The point is not the mere fact that the Netherworld is feared
by the gods, but that it is so dreadful that it is feared even by the gods.The constituent 'the gods' {theoi) is used in (1) in a way which is highly
characteristic of Homeric, or, generally, Archaic Greek thought: the gods arenot referred to for their own sake; they are mentioned in relation to 'man'. InHomer, 'god' and 'man' may be seen as a binary pair whose contrast is con
stituted by the presence or absence of immortality. As immortality is naturally associated with 'strength' and 'courage', it is not surprising that 'god'repeatedly implies 'stronger than man'. Consequently, someone who commitshimself to the truth of 'The gods are scared of x' will a fortiori commit himselfto the truth of 'Men are scared of x'.
This piece of the Homeric world view has an interesting linguistic implication, in that it partly determines the way the substantives 'gods' and 'men' may
be used in the language. Gods and men can be conceived of as items on a scalein which 'gods' represents the high point and 'men' its low point, or, better, a
lower point. Thus we may say that 'the gods' in (1) does not refer to the godsqua gods, but to the gods qua (highest) item on the following scale:
(2)
Scalarity is a semantic notion which applies when entities or propertiescan be ordered on a scale, like gods, men. We may say that between 'gods' and'men' a scalar relationship obtains. Gods have strength, power and immortalityand these properties are frequently measured against the weakness and mor
tality of man. Accordingly, the items on the scale in (2) can be ordered by degree of decreasing strength. Scalarity and scalar relationships can be encoded
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 39/318
THE LINGUISTIC BACKGROUND: SCALARITY 29in language by a number of 'scalarity markers'. It is with one such a markerwhich is peculiar to Homeric Greek that the present study is concerned.2
Scalar relationships are a kind of paradigmatic relation; they involve therelation between a given item and the item(s) that may be substituted for it.(Of course, when we say that two or more items may be substituted for oneanother, we allow for differences in meaning between those items.) Paradigmatic relations may be opposed to syntagmatic relations, which obtain betweenitems belonging to one and the same structure (sentence, noun phrase etc.).
Thus between 'Greek' in 'the Greek language' and 'Latin' a paradigmatic relation is contracted, whereas between 'Greek' and 'language' we have a syntagmatic relationship.
Sometimes it suffices for a paradigmatic relation between two items toexist that both items are of the same grammatical category. In the case of scalar
paradigmatic relationships, however, there is a constraint on substitutability inthat the substituting item has to share, in addition, a semantic feature with theitem substituted for, so that they can be placed on one and the same scale. Thissemantic feature they have in common is precisely that which is measured inthe scale. The notion of substitution is important and will recur later on (see2.3.1.3 below).
The term 'scale' can be dealt with in linguistics in various ways. First, inPrototype Theory as discussed in 1.4, 'scale' may be used in the sense of a continuum between two extremes. The difference between this kind of scale andthe one exemplified by (2) is that in (2) there is only one extreme, all othervalues being merely 'lower'. Second, we can make use of 'scales' in connectionwith the notion of 'implicature'. In this approach, which is not practiced in the present study (but see 2.3.1.1 below), the notions of scale and substitution areused to account for certain aspects of (co-operative) language-behavior.
Third, scalarity may be used when one is dealing with 'at least' and 'at most 9
meanings. Language is often scalar in the sense that a given item 'q', which occupies an intermediate position on a scale , q, r may allow of the readings 'atleast q' ('q, if not p') or 'at most q' ('r in any case, but perhaps q') beside theliteral, 'non-scalar' reading ('q'). This kind of scalarity may involve the study of
particles like only, but, merely and their equivalents in the various languages.In the present study particles of this class will be called focus particles (see further 2.3 below).
These three senses/uses of the word 'scale' are concerned with a scale asa whole; in using 'scale' in one of the three senses one is equally interested inthe extreme value of the scale and in the lower values. Sometimes one even as-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 40/318
30 LINGUISTICS AND FORMULAS IN HOMERsumes two extremes. When using 'scale' as I do in the present chapter, on theother hand, we are only interested in the high point of a scale, all the othervalues having no identity other than being lower'.6 The scales we are concernedwith are imequivocally structured and oriented; they run downwards in onedirection from 'strong' to 'weak', or generally: from an extreme, a superlativevalue to lower values.
Just as the scales associated with at least and at most meanings, the 'unique high point scale' can be discussed in connection with particles. These are
the kind of particles that are typically concerned with the high point of the scale.The kind of particles that is associated with scales like (2) I will call scalar particles (as opposed to mere focus particles). The obvious example of a scalar particle is even and its equivalents in other languages.
2.1.2 Superlatives: scalar and attributive
One of the essential properties of scales as I have defined them is super-lativeness. The high point of a scale has, by definition, superlative value. It functions as the high point by virtue of that very value. In English, a typical way to
mark scalarity is by means of superlatives. Consider:(3) He could solve the most complex problem.
(4) He couldn't solve the simplest problem.
Ex. (3) could be taken to mean that the person in question was able, at a particular occasion, to solve a particular problem, which was more complex thanany other problem in the context referred to. Likewise, (4) could be taken tomean literally that the person referred to was able, in a particular context, tosolve all the problems present, except the simplest one. In this reading of (3)-
(4), in which the propositions are taken to refer to a particular state of affairs8in the past, the superlatives function as simple, referential noun phrases. Theyrefer to a particular problem, which is (perhaps accidentally) either the mostcomplex or the simplest one in the situation in question.9
A more natural reading of (3)-(4), however, is the one in which scalarityis involved. In this reading (3) is taken to mean that the person in question isso clever as to be able to solve any problem presented to him, and (4) that heis so stupid as to be unable to solve any problem. In this reading, the superlative refers, by virtue of its superlative value, to the high point of a scale. As a
result, it has implicative force: someone who can solve the most complex problem can solve, by implication, all less complex problems. Likewise, some-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 41/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 42/318
32 LINGUISTICS AND FORMULAS IN HOMER
simplest (problem)' in (4) is equivalent to any as an existential quantifier, theexistence of problems the person in question is able to solve being negated.
The similarity of scalar superlatives and any is in itself not very importantfor the purposes of the present study. It is, however, useful in a discussion ofthe peculiar behavior of scalar superlatives with regard to negation. In section2.4 below I will present this discussion (see also 3.4. below).
Not only grammatical (morphosyntactic) superlatives allow of a scalar('quantified') reading. Pragmatic' superlatives may occur in scalar statements,
too (cf. Fauconnier 1975a: 353), as appears from(5) Kasparov couldn't win this position.
(6) Paul Getty would refuse to buy this painting.
In the preferred reading these examples are not simple assertions about the persons referred to. 'Kasparov' and 'Getty' are not meant to be referentialterms here; they are meant to represent the high points of the scales of chess
players and millionaires, respectively. Consequently, (5)-(6) are not in the first place propositions conveying information that is important for its own sake. In
their scalar reading they are uttered to support (viz. function as an argumentfor) another, unexpressed, proposition which refers to a 'superlative' state ofaffairs (see 2.1.2 above), for example:
(5)' This position is impossible to win: no one could win it.
(6)' This painting is intolerably expensive: anyone would refuse to buyit.
2.1.4 Irrealis modality
In the present subsection I discuss another important property of scalarsuperlatives. The distribution of scalar superlatives is constrained in a remarkable way: just as superlatives when used as scalar superlatives cannot bereferential, so they cannot occur in sentences with what is commonly calledspecific temporal reference. It will appear that both properties are related. Thedistributional constraint is not, as is often the case in linguistics, a general tendency (see 1.4 above), but an exceptionless rule.
The constraint upon the distribution of scalar superlatives can be best discussed in terms of the modality of the predicate of the sentence in which thesuperlative occurs. In this text I will use 'modality' in the broad sense the termis given in the work of Givón (e.g 1973, 1979, 1984). In this sense modality is
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 43/318
THE LINGUISTIC BACKGROUND: SCALARITY 33the axis along which propositions are factual or non-factual. In the first case wemay speak of realis modality and in the second case of irrealis modality. Realismodality is concomitant with specific temporal reference: the proposition inquestion refers to an actually obtaining ('existing') state of affairs which is
bounded in time and space. In the case of irrealis modality, on the other hand,specific temporal reference is absent. The state of affairs to which the proposition refers either has no (clear) boundaries along the time-axis or it does notobtain.
It will be clear that 'modality' in this sense differs from the term as it isused in the logical tradition. In logic and in logic-based linguistics 'modal' isfrequently opposed to 'non-modal', the former being characterized by conceptssuch as 'necessity' and 'possibility' and the latter by the absence of these notions. In the present study, on the other hand, modality is not conceived of inthis binary way, in which the 'modal' is marked with respect to the 'non-modal'.All verbal words, regardless of whether they are morphosyntactically markedfor 'mood', are equally regarded as 'modal', some of them being factual andothers non-factual. This strategy has considerable advantages in descriptive linguistics.
Modality is intimately bound up with the tense-marking of the verbalword. Factual sentences tend to have predicates with present or past tense-marking, whereas future predications are typically non-factual. The ontologi-cal basis of this tendency is clear: what is actually happening, or what actuallyhappened can be naturally conceived of as 'real', 'factual' ; what has not yet hap
pened, on the other hand, is as yet 'irreal'. Future states of affairs do not (yet)exist, by definition. Consequently, their typical modal character is non-factual.17
Beside sentences with overt future tense marking irrealis modality can be
marked in many ways. Many cases of irrealis modality have the same ontologi-cal basis as the future in that the state of affairs expressed by the propositiondoes not yet exist factually and undeniably. In this connection we have to mention the non-indicative verbal moods (expressing uncertainty or probabilityrather than certainty), deontic modal verbs and the complements of verbslike 'want', 'plan', 'try', 'strive', etc.20
Of special interest for the present study (see 3.1.1, 3.1.2 below) is themodal character of the generic (or 'habitual') present tense. The generic
present is a present tense, and qua verb form with present tense marking it may
be used for factual statements; the state of affairs to which a generic proposi-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 44/318
34 LINGUISTICS AND FORMULAS IN HOMERtion refers exists either permanently (c.q. during a considerable stretch of time)or intermittently. Consider the following examples:
(7) Apollo scares the bravest warrior.
(8) Whenever I visit John, he is reading a book
These sentences are clear, straightforward, assertions of fact. But this observation does not exhaust their semantic properties. The states of affairs to which(7)-(8) refer have no real boundaries along the time-axis; their truth does notdepend on any particular circumstances. This is different in:
(9) Apollo scared the brave warrior by his fierce appearance.
(10) When I visited John last night, he was reading a book.
The difference between these sentences and (7)-(8) Ues in the specificness ofthe temporal reference. Exx. (9)-(10) refer to a particular state of affairs; thisis why we call them factual. Exx. (7)-(8), on the other hand, have a mixed statuswith regard to modality: they are factual in so far as the state of affairs to which
they refer actually exists, but to the extent that they lack specific temporalreference they are non-factual.On account of its non-specific character, the generic present may often be
conveniently ranked among the irrealis phenomena, in spite of its being thestatement of a fact. This treatment yields a number of insights that cannot bearrived at when we conceive of the generic present as a purely factual affair. Inthe present study I discuss the advantages for the description of scalar
phenomena; elsewhere (Bakker 1988a) I have tried to make the insight of thegeneric present as an irrealis phenomenon productive for the unified description of the subjunctive conditional in Ancient Greek.
There is a systematic connection between the modality of the predicateand the referentiality of the nouns or noun phrases that function as its arguments. In other words: spatial reference is closely connected with temporalreference. The most straightforward and common example of this
phenomenon is the referentiality of indefinite noun phrases.23 These noun phrases tend to have a referential understanding when they function as arguments of a verb which is marked for realis modality. When they are under thescope of an irrealis modality, on the other hand, they are ambiguous ('referentially opaque'), though preferably understood non-referentially. For illustra
tion of this, look at the indefinite noun phrase a book in (8) and (10). In (10),a realis environment, it refers to a specific, particular book (the speaker may
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 45/318
THE LINGUISTIC BACKGROUND: SCALARITY 35
or may not know its title), whereas in the case of (8) it is preferably understood as not referring to any particular book. The reading of (8) in which Johnis always reading one and the same particular book when the speaker enters issomewhat strained.
We now return to the non-referentiality of scalar superlatives, which wasdiscussed in 2.1.3. Here, too, the connection between referentiality andmodality obtains. We just saw that indefinite noun phrases may be referentialor non-referential, according to the modality of the sentence in which they
occur. Scalar superlatives, on the other hand, are necessarily, categorially non-referential. We may now add that it follows from this that they are distribution-ally confined to irrealis environments. See, for instance, exx. (3)-(4) above. Onaccount of the modal verb (could) these sentences are non-factual environments where specific temporal reference is absent. The same applies to (5)-(6): the scalar possibilities of the proper names 'Kasparov' and 'Getty' comeinto play by virtue of the non-factual environment. They disappear when an unequivocally factual predicate is introduced (viz. when the modal verbscould/would are deleted).
Scalar superlatives are, In English, not overtly marked as such; they ap
pear as morphosyntactic superlatives. Thus we have to say that when their non-factual environment is turned into a factual one, their scalar meaningdisappears, whereas their form remains. Similarly, an indefinite noun phrase isturned from a non-referential into a referential one when the sentence in whichit occurs is turned from an irrealis environment into a realis environment. InHomeric Greek, on the other hand, scalar superlatives are overtly marked assuch, by per as we shall see (3.1.1). It follows from this that irrealis modality
puts a constraint on the distribution of per when it functions to mark a scalarsuperlative. Thus, whereas in English non-factual modality is a necessary con
dition for the scalar reading of a superlative, in Homeric Greek it is a necessary condition for the occurrence at all of a scalar superlative.
In fact the constraint put upon scalar (readings of) superlatives bymodality can be gauged from the similarity of scalar superlatives and theEnglish quantifier any (see 2.1.3 above).Any as an 'existential' quantifier is dis-tributionally restricted to negated propositions, conditional clauses and yes/noquestions. The non-factual (non-declarative) nature of all these contexts, whichare often called 'polarity contexts' (see 2.4.2 below) is apparent. Negationmay, under certain conditions, be treated as a non-fact modality. Conditionals,too, are a typical case of non-fact modality, on account of their hypothetical
character. Likewise, 'universal' any occurs in non-factual environments
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 46/318
36 LINGUISTICS AND FORMULAS IN HOMER
without specific temporal reference: contexts containing modal verbs {can,could, will, would, see exx. (3)-(6) above),28 the generic present and imperatives.29
2.2 Scalarity and referentiality
In the previous section we saw that grammatical (morphosyntactic) and pragmatic superlatives may allow of a scalar reading. We also saw that there isan important distributional constraint upon this phenomenon, which is due to
irrealis modality. But scalarity is by no means confined to scalar superlatives.There are other markers of scalarity, whose distribution is not constrained likethat of scalar superlatives. One such a marker is the particle even in English.The present section (2.2) provides the link between scalar superlatives (2.1)and the discussion of the semantics of even (2.3). Its subject matter providesthe appropriate context for the introduction of a distinction which plays acentral role in the present study: the distinction between 'extension' and'intension', to be discussed in 2.2.2.
2.2.1 Attributive and referential superlativesThe superlatives discussed in the previous section allow of a scalar read
ing by virtue of their very meaning, that is, simply, by virtue of their being a su perlative. Let us pursue the consequences of this observation.
'Superlativeness' is, morphosyntactically, a category exclusively applyingto the adjectival word. From the point of view of ontology, only properties ofentities ('things', persons) can be subjected to superlativeness, degree and com
parison, not the entities themselves. And it is properties that are typically encoded in languages by means of adjectives.30
Like all other adjectives and terms with descriptive content, superlativescan be used to ascribe (attribute) some property to some entity. In using thesuperlative, one may assert that a given property applies to a given person orthing to the highest degree in the context of utterance:
(11) Bill is Peter's best friend.
This example may be interpreted as an 'ascriptive sentence' (see Lyons (1977:148, 185)) in which a certain property, expressed as a non-verbal predicate, isattributed to the referent of some subject-term. In (11) it is the property of
being 'Peter's best friend' that is attributed to Bill; accordingly, the (noun) phrase to which the superlative belongs is interpreted as non-referential.31 It
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 47/318
THE LINGUISTIC BACKGROUND: SCALARITY 37
has, in saying something about , an ascriptive, or descriptive ('attributive')role.32
Now Teter 's best friend', which is used attributively in (11), may be usedreferentially, especially when it functions as subject:
(12) Peter's best friend died last week. (That's why Peter is sodistraught.)
Given the fact that Peter's best friend is (was) in fact Bill (on the strength of
(11)), the proper name Bill may be substituted for Peter's best friend in (12) withthe truth-conditions of that sentence remaining unchanged: (12) remains astatement about the same person. (This is called 'Leibniz's principle of substitution of co-extensive terms salva veritate'. However, although this substitution does not turn (12) into a false sentence, it does very probably turn (12) intoan inappropriate (or infelicitous) one. The point of (12) as it stands is to providean explanation of Peter's afflicted appearance. In referring to Bill the way hedoes, the speaker of (12) assumes that his addressee is not acquaintained withthe truth of (11). And as being Teter's best friend' is the crucial attribute ofBill in the context of utterance of (12), the speaker refers to Bill in that way.
332.2.2 Intension and extension
The simple, obvious, yet fundamental insight to be gained from (11)-(12)is that things or persons may be referred to in various ways, according to thecircumstances in the context of utterance (e.g. the pragmatic knowledge theaddressee supposedly has, or lacks). Thus, whereas Peter's best friend in (11) inthe ascriptive reading is only attributive (property ascribing), in (12) it has twofunctions: it refers to (picks out from the multitude of possible referents) therequired person, but at the same time it supplies the relevant A r tdesGegeben
seins (Frege 1892).34
More than one description may be true of one and the same object andany object can be referred to in more than one way. Let us for the moment dis
pose of the pragmatic implications of this observation and focus on the merefact that Peter's best friend in (12) has a double function. The two functions area reflection of two aspects of noun phrases. Noun phrases may have a 'pointerfunction': they point to the person or thing a given language user wants to referto. This person or thing may be called the extension of the noun phrase. In(12) the extension of Peter's best friend is the person named 'Bill'. But at the
same time noun phrases may be descriptive. This is called their intensionalaspect. The intension of referential noun phrases is what they 'say' about their
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 48/318
38 LINGUISTICS AND FORMULAS IN HOMERAs we saw, the intension, i.e. the choice of a particular aspect of the referent,is often, if not always, determined by pragmatic factors.
The distinction between intension and extension is an old one in philosophy; it has been dealt with in various terminologies, and it has obviousimplications for logical theory.38 In the present text, the distinction will bemainly used as a basis for the construction of a continuum along which noun
phrases can be ordered by degree of intensionality (descriptive content). Thiscontinuum involves a scale of the type that is used in Prototype Theory (see
1.4), one with two extremes, with continuous space in between.On the one extreme of the scale we place terms with a maximum amountof extensionality and a minimum amount of intensionality, the terms that havemost clearly a 'pointer function'. The obvious candidates for location at thisend of the scale are proper names as well as demonstrative pronouns and otherdeictic elements. Proper names are essentially mere 'labels' for their referents.Speakers normally do (need) not say anything about a person whom he refersto with a proper name. At the other extreme of the scale we find opposite
phenomena, terms with a minimum amount of extensionality and a maximumamount of intensionality, the terms that are preferably used to describe(characterize) a person or thing, rather than refer to him/her/it. The examplethat presents itself here is the kind of noun phrase on which superlativenessoperates, viz. the adjective. The prototypical adjective is purely descriptive,
property ascribing; it does not refer itself. In between the extremes we have tolocate, with as many gradations as one is prepared to make, noun phrases like'Peter's best friend' in (12), in short: all ordinary simple nouns and compositenoun phrases.
The continuum from extensional to intensional is one of descriptive content, viz. an inherent, context-independent property of noun phrases. Conse
quently, the distinction between 'intensional' and 'extensional' is different fromthe distinction between 'referential' and 'non-referential'. For referentiality ofnoun phrases is not an inherent property; it is a matter of how they are used.Intensional terms with a high degree of descriptive content can be perfectlyused to refer to something ('Which one do you want? - The red one'). And,conversely, extensional terms with a low degree of descriptive content can beused to describe/characterize someone, rather than refer to him/her. Someyoung and very promising chess player, for example, may be said to be 'aKasparov', or 'a new Kasparov'. And by the same token extensional terms can
be used as scalar superlatives (see (5)-(6) above).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 49/318
THE LINGUISTIC BACKGROUND: SCALARITY 39 Notice,finally,that there is a clear relationship between intensionality and
predicates: the terms that are likely to function as (time-stable) predicates haveto be located on the intensional part of the scale. Thus we may say 'Mary is
beautiful' but not 'Mary is Suzan', unless we are engaged in a plot of mistakenidentity. But then we are equating, not predicating.
2.2.3 Scalarity and extensional terms
We now return to scalarity and consider it against the background of the
intension-extension distinction. In section 2.1.1 and 2.1.21 stated that superlatives may allow of a scalar reading. When they do, they are (meant to be) non-referential, just Hke nouns or noun phrases modified by the quantifier any. Theyallow of their scalar reading by virtue of their meaning, that is, by virtue of their
being a superlative. We can now characterize 'superlativeness' as a wholly intensional affair. In other words, scalar superlatives have to be located at the intensional extreme of the intension-extension continuum.
But scalarity is not confined to intensional terms. Extensional terms, too,may be used in a scalar way. However, as extensional terms are devoid of
descriptive content, they are unable to do so on their own. Someone who wantsto make a scalar statement in uttering (13) below very probably will not makeclear his communicative intentions:
(13) Paul liked the excursion.
'Paul' is a purely extensional term without descriptive content, whereasscalarity as a paradigmatic relation between two items (see 2.1.1) presupposessome property by degree of which the two items can be ordered. Thus, whenmeant as a scalar assertion, (13) may well be a communicative failure. It willnormally be understood as a simple, non-scalar, statement about Paul. If aspeaker is to express his scalar intentions in a clear, unambiguous way, he hasto modify the extensional term by the scalar particle even:
(14) Even Paul liked the excursion,
thereby communicating that Paul was the least likely person to approve of theexcursion in question. Thus the element conveying scalarity, which is potentially present in any intensional term, has to be expHcitly added in the case of extensional terms in the form of a scalar particle. It is to this kind of particle, andthe way it can be described, that we must now pay attention.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 50/318
40 LINGUISTICS AND FORMULAS IN HOMER
2.3. Focus particles and focus contexts
The particle even, which may be used as a marker of 'extensional scalarity'(scalar relations between terms without descriptional content), may be said to
belong to the class of focus particles. Sometimes (e.g. König 1981), this class of particles is called 'scalar particles'. Yet I prefer 'focus particles' because 'scalar'suggests a connection between the class as a whole and the scalar phenomenadiscussed in the previous section. But this connection exists only in the case ofeven. Thus even may be called a 'scalar focus particle', the scalar particles being
a subclass of the focus particles.Focus particles, whether scalar or not, are always accompanied by a con
stituent which may be called their focus constituent (see 2.3.1.3). The focus constituent contracts a paradigmatic relation with the alternatives which may besubstituted for it. Furthermore, focus constituents always attract the main stressof the sentence. In subsection 2.3.1 I discuss the particle even as a focus particle, with due attention to the scalar properties of the particle. In 2.3.2 I willdiscuss focus particles in a wider perspective, in connection with the conceptof 'focus context'.
2.3.1 Even as a focus particle
2.3.1.1 Conventional implicature and the meaning of even
Much has been written on the subject of the contribution of even to themeaning of the sentence in which it occurs. In spite of many differences inmethod and outlook, most authors seem to agree that the nature of this contribution cannot be characterized as 'truth-conditional'.40 In other words, theyagree that the 'truth-conditions' of a sentence containing even (or any otherfocus particle) are identical to those of the same sentence without the particle.
Thus exx. (13) and (14) above are said to refer to one and the same state of affairs. But if the meaning of even is not truth-conditional, how are we to accountfor it? In older literature (e.g. Horn 1969), the meaning of even, also/too, andonly is dealt with in terms of presupposition: in uttering ex. (14) above we assert
the fact that Paul liked the excursion (i.e. the truth-conditional part), and we presuppose that others did so too.
Apart from the fact that in uttering (14) we convey just a little more (viz.that Paul was the least likely person to like the excursion in question, see below),it remains to be seen whether the non-truth-conditional part of the meaning of(14), i.e. the meaning of even, has to be dealt with in terms of 'presupposition'.
The term 'presupposition' may be used to refer to what is 'given', 'old' infor-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 51/318
THE LINGUISTIC BACKGROUND: SCALARITY 41
mation in a proposition that is part of a coherent discourse (e.g. in Givón1979). And in a more technical sense the term is used in Fregean/Strawsonianlogical semantics to designate what is considered in this approach an inherent
property of propositions and definite noun phrases. However, the term'presupposition' gradually became overburdened. As Karttunen & Peters(1979: 2) put it, "a wide range of different things have been lumped togetherunder this single label and this fact is, more than anything else, responsible forthe continuing controversy about how to analyze presuppositions."
Karttunen & Peters suggest that non-truth-conditional phenomena betreated under more than one label. The appropriate label for the meaning ofeven (and similar elements), they argue, is the notion of conventional implica-ture. In thepresent section I follow this approach, partly in want of a better alternative. Karttunen & Peter's analysis has influenced the importantdiscussion of focus ('scalar') particles in König(1981).
The notion of 'implicature' stems from Grice (1975). By 'implicature'Grice means anything that is 'implicated' by a speaker by uttering a sentenceor a constituent. We implicate something when we allow our hearer or addressee to infer something from our utterance over and above what we have actually, literally, said. Implicatures may either conventional or non-conventional.
Non-conventional (or 'conversational' in Gricean terms) implicatures are entirely bound up with a particular context or situation. They are ultimately connected with what Grice calls the 'cooperative principle': what is actually saidin a particular situation will normally be interpreted by a hearer as uttered witha cooperative intention. The interplay of what is said with what is expected onthe part of the hearer on the basis of the cooperative principle gives rise to im
plicatures.
Conventional impHcatures, on the other hand, are connected with the con
ventional meanings of words and other elements of the language system. A certain word or phrase or construction may implicate something by its verymeaning. The kind of meaning connected with this conventional implicaturehas to be kept separate from the literal ('truth-conditional') meaning of thesentence in which the implicature arises.
Now what is the conventional implicature even gives rise to? In Karttunen& Peters (1979:12) it is stated that someone uttering a sentence of the type
(14) Even Paul liked the excursion
commits himself to the truth (I would say 'relevance') of three propositions:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 52/318
42 LINGUISTICS AND FORMULAS IN HOMER
(15) A. Paul liked the excursion.B. Other people in addition to Paul liked the excursion.C. Of the people under consideration, viz. the participants of theexcursion, Paul was the least likely to like the excursion.
These three propositions do not have one and the same status, as Karttunen &Peters point out. (15)A is what the speaker of (14) actually says with so manywords; it represents the 'truth-conditional' part of the meaning of (14), or, alternatively, the propositional content of (14). (15)B and C, on the other hand,are not actually asserted by the speaker of (14); these propositions are conventional implicatures arising from the presence of even in (14). These implica-tures do not arise involuntarily; they are closely connected with thecommunicative intentions of the speaker of (14). If the speaker of (14) did notwant to commit himself to the truth of (15)B-C, he would not have used evenwhen uttering (14).
Karttunen & Peters (followed by König 1981) argue that there are twoways in which the difference between 'truth-conditional' (asserted) and implicated meaning can be brought out. The first way consists in embedding ex. (14)
under a factive predicate:(16) I noticed that even Paul liked the excursion.
What the speaker noticed, Karttunen & Peters argue, is the truth of (15)A, thefact that Paul liked the excursion (i.e. the truth-conditional part of the meaning of (14), not the fact that there were other people who did so or the fact thatPaul was the least likely person to do so, i.e. the implicated part of the meaning of (14). (However, this account does not do full justice to the meaning of(16), see the following subsection.)
Karttunen & Peters' second way to bring out the difference consists in bringing (14) under the scope of if, i.e. in treating (14) as a conditional:
(17) If even Paul liked the excursion, then it was a very interesting one.
Here the truth-conditions of (14) are suspended, just as in the case of allhypothesized propositions. (Note that conditionals are, in the terms of section2.1.4 above, a typical non-factual (irrealis) environment. ) Consequently, aspeaker who commits himself to the truth of (17), i.e. to the validity of the im
plication expressed by it, will not (normally) commit himself to the truth of(15) A. However, he does commit himself to the truth of (15)B-C. His doing so
is even the very point of the main clause of (17). In other words, the implica-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 53/318
THE LINGUISTIC BACKGROUND: SCALARITY 43tures arising from the presence of even in (14) remain intact; they are notsuspended by the hypothesis.
2.3.1.2 Elocution and the use of even
Thus we see that there are some grounds for distinguishing between the'truth-conditional' part of the meaning of a sentence like (14) and the specificcontribution of even to that meaning. However, the description of even and thesentences in which it occurs is not finished with this distinction. In being ex
clusively concerned with the (context-independent) meaning of sentences,48 itdoes no justice to the intention which speakers have when uttering a sentencein which even occurs. This neglect of the pragmatic component of the use ofeven may be considered a serious shortcoming of Karttunen & Peters' account(as well as of that of other authors).
It was stated that a speaker uttering a sentence Hke (14) above commitshimself to three propositions. But this commitment is not the reason for his uttering (14): if he wanted to assert the truth of (15)A, he did not have to useeven, and if he wanted to assert (15)B or (15)C, he simply could have asserted
these propositions, instead of implicating them by even. In that case, even neednot have been used either.The point is that even with its conventional implicatures is used as an in
direct assertion or as an illustration of some remarkable and superlative fact.The point of uttering (14) may be the assertion, in an indirect way, of the factthat the excursion in question was a very interesting one, given the fact that itwas enjoyed by the least likely person to approve of it. And in (16) the 'superlative fact' is what the speaker really noticed.
Thus the communicative potential of sentences with even is similar tothose containing scalar superlatives (see 2.1.2). What is expressed in the caseof scalar superlatives by the descriptive content of the superlative, is expressedin the present case by the meaning (implicature) of even. Both scalar superlatives and even may be used for the assertion of superlative facts. On accountof this use we may say that both even and the scalar superlative determine tosome extent the illocutionary potential of the sentence in which they occur. By'illocution' we simply mean what is actually done with a given sentence in agiven situation, the kind of speech-act that is performed with it.51 Sentencescontaining scalar expressions usually have an illocutionary force which does notcoincide with the simple assertion of the 'truth-conditional' part. Thus the im
plicature of even is not something to which one commits oneself for its own
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 54/318
44 LINGUISTICS AND FORMULAS IN HOMER
sake; it is ultimately subservient to the illocution of the sentence in which evenoccurs.
The notion of 'superlative fact' will be used in this study as a technicalterm, to designate, in connection with the notions of speaker intention and il-locution, the reason for, and the explanation of, the presence of a given scalarexpression in a given sentence. In section 3.2.21 will further discuss the typicalillocution of sentences containing even.
2.3.1.3 Focus and scopeEvery scalar particle is associated in the sentence with a constituent that
can be called the focus constituent of the particle.52 The focus constituent('Paul' in (14)) always attracts the main stress of the sentence.53 In connectionwith the notion of 'focus', the first implicature of even ( = (15)B) can now beformulated as follows. Even implicates that the sentence in which it occurs isalso xrue of entities other than the referent of the focus constituent. In otherwords, it is implicated that there are alternative values for the focus constituent.Between these alternative values and the focus constituent a paradigmatic rela
tion is contracted (see 2.1.1). On this account, even (just as also and too) ranksas an inclusive focus particle. A particle like only, on the other hand, is an exclusive focus particle: it implicates that the sentence in which it occurs is trueof no one or nothing other than the referent of its focus constituent, alternative values being excluded. For further discussion and application of the distinction between inclusive and exclusive focus particles see 3.5.2 below.
The distinction between 'inclusive' and 'exclusive' focus particles can also be stated in terms of substitution. In the case of inclusive focus particles, alternative values may be substituted for the focus constituent, without the 'truth-conditions' of the sentence changing. In the case of exclusive focus particles,
on the other hand, this is impossible. The idea of substitution leads to thesecond important notion in connection with focus particles, their scope. 'Scope'we define (again following Karttunen & Peters 1979) as an open sentence witha variable in the place of the focus constituent of the particle. We can get thescope of a given particle by deleting the particle and replacing its focus constituent by the variable. Thus, the scope of even in ex. (14) is:
(18) X liked the excursion.
The scope sentence of an inclusive focus particle is the open sentence which is
'true' of the focus constituent as well as of the alternative values for the focusconstituent. Thus the meaning (impHcature) of even depends both on the focus
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 55/318
THE LINGUISTIC BACKGROUND: SCALARITY 45
and on the scope of the particle. The scope puts constraints on the alternativesof the focus term. For example, in (14), when Bill was not a participant of theexcursion in question, he is ruled out as an alternative for the focus constituent.
Even has inclusiveness in common with abo and too. But there is also animportant difference. Unlike also and too, even imposes an ordering upon thefocus term and its alternatives: this ordering is one of increasing likelihood.The focus constituent of even typically represents the high point on a scale, allalternative values representing points that are lower on the same scale. Just as
in the case of scalar superlatives, the scale associated with even is the kind ofscale of which only the high point matters. The extensions (see 2.2.2) of the possible alternatives for the focus constituent of even are more likely to conform to the truth of the scope sentence, and in representing a non-superlativevalue they are less important than the focus constituent itself.
We are here dealing, of course, with the second implicature associatedwith even (ex. (15C above), which Karttunen & Peters rightly call 'scalar'. Evennot only implicates that there exist alternatives for its focus constituent, butalso that those alternatives are less extreme cases than the extension of thefocus constituent. The scope sentence determines with respect to what they areless extreme.
2.3.1.4 Scope dependence and scope independence
The ordering of the focus constituent and its alternatives, in other words,the scale whose validity is implicated by a scalar focus particle, may be eitherdependent or independent of the scope sentence of the particle (König 1981:119). This distinction should be studied in close connection with the distinction
between 'intension' and 'extension' (section 2.2.2 above). A scale whose high point is represented by an intensional term is scope- (i.e. context-) inde
pendent, whereas scales whose high point is represented by an extensional termare usually dependent on their scope ('context'). Accordingly, scales associatedwith scalar superlatives are scope-independent: the scale exists by virtue of themeaning of the superlative. For example, it is clear a prion that anyone who cansolve 'the most complex problem' will a fortiori be able to solve less complex
problems, for 'the most complex' implies 'less complex' independently of scopeand context, all complex things being by definition less complex than what is'the most complex'. In (14), on the other hand, it does by no means follow fromthe meaning of the word 'Paul' (if an extensional term has any meaning at all)
that this person was the least likely to enjoy the excursion. Accordingly, the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 56/318
46 LINGUISTICS AND FORMULAS IN HOMERscale whose high point 'Paul' represents is dependent on the scope sentenceand of the context.
To convey (viz. implicate) the existence of a scale of which an extensionalterm represents the high point, the particle even has to be introduced, inEnglish. Thus, when even takes an extensional term as its focus constituent, itcan be called a scope dependent focus particle (which is shorthand for 'focus
particle whose implicated scale is scope-dependent'). However, as English is alanguage with very few scalar particles, even has to perform a multitude of func
tions. It has to do the job of scope-independent particles as well, for example:(19) Even elementary Greek grammar is too difficult for him (let alone
advanced reading).
The scale on which knowledge of elementary grammatical facts and ability toread any text in a language are situated is scope independent. In languageswhich are richer in scalar particles than English, e.g. German and Dutch, evenin (19) may be translated by another particle than even in (14). Scope-dependent even in (14) is equivalent with sogar/selbst (German) and zelfs (Dutch).Scope-independent even in (19), on the other hand, is equivalent with schon
(German) and nog (placed after the focus constituent) or alleen al in Dutch.We shall see that in Greek, too, a distinction may be made between scope-de
pendence and scope-independence: the combination of the conceptual pair 'intension-extension' with scope dependence and will prove useful for thediscussion of per and kai in chapter 3.
2.3.1.5 Scope ambiguity
Even in EngUsh is an overworked particle; this situation gives rise to am biguities which do not exist in German and Dutch, because these languageshave more scalar particles, and. hence, a better division of labour.57 Even may
be ambiguous as to its scope. This situation occurs when even is found in atype of context which may be exemplified as follows:
(20) I don't think that he understands even Homer.
(21) If he can read even Homer, we will praise him.
These sentences are ambiguous as to the scope of even: in (20), this scope will be preferably understood as 1 don't think that he understands X. On this read
ing, which we
call 'wide scope', the intended meaning of (20)
is that Homer iseasy to understand (Karttunen & Peters (1979: 27)): if we don't think that
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 57/318
THE LINGUISTIC BACKGROUND: SCALARITY 47
someone understands even Homer ( = 'if we don't even think someone understands Homer'), we will consider this person even less capable of understandingother (i.e. more difficult) authors. However, there is another reading, in which(20) implicates just the opposite (viz. that Homer is difficult to understand).This is the narrow scope reading of (20), in which the scope sentence of evencoincides with the embedded sentence He understands X. Notice that the am
biguity is due to the negation: when we delete don't, the ambiguity disappearsand only the narrow scope reading remains.
Exactly the same applies to (21). In the narrow scope reading, the scopeof even is he can readX, that is, the sentence that functions as the conditionalin (21). In the wide scope reading of (21), on the other hand, the scope sentence of even comprises the conditional expression as a whole, that is, the conditional plus its main clause: If he can read X, we will praise him. In the narrowscope reading the following paraphrase is possible: 'If < it is true that > [hecan read x], then < it is true that >...'. In other words, the scope sentence (insquare brackets) is part of a conditional construction; it fills the slot 'p' in If p,then q. In this reading, the clause functioning as conditional ('p') may be seenas a proposition in its own right. In the wide scope reading, on the other hand,the conditional and its main clause together form one complex sentence, theconditional stating the circumstances under which the person in question will
be praised.61
In (21), too, the scope-difference is connected with a difference in implica-ture. On the narrow scope reading we implicate that Homer is a a difficultauthor, whereas on the wide scope reading he is implicated to be very easy tounderstand. This difference in implicature arises because the wide scope sentence and the narrow scope sentence are associated with two qualitatively different scales. In the scale associated with narrow scope, 'Homer' represents a
high point with a positive superlative value. This is the kind of scale which has been discussed hitherto. In the scale associated with wide scope, on the otherhand, 'Homer' represents an extreme point with negative (or 'inverted') superlative value. The distinction between 'positive' and 'negative' is not confinedto even. It occurs, too, in the case of the quantifier any and scalar superlatives.In section 2.4.2 below I will discuss the phenomenon in its own right, in connection with the notion of 'negative polarity'.
It is worth pointing out that the scope ambiguity in examples like (20) and(21) cannot arise in Dutch and German. These languages use different particles
for different scopes, and hence for different implicatures: auch nur
(G.) andook maar (D.) translate wide scope even, whereas narrow scope even is
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 58/318
48 LINGUISTICS AND FORMULAS IN HOMER
equivalent to sogar/zelfs. However, the EngUsh language partly compensatesfor the lack of variety in its array of scalar (focus) particles: often special negative (inverted) superlatives like (the) slightest do the job of the Dutch and German wide scope particles. See further 2.4.2 as well as 3.4.3 below.
2.3.1.6 Even if: concession and concessive conditionals
In the previous subsection I have discussed the conditional clause in connection with the scope of even. We saw that a conditional clause may function
as the scope of even or as part of it. But it may equally function as the focus ofthe particle. In that case we have the even //"-conditional, which is often referredto as the 'concessive conditional.'62 Chapter 6 will be devoted to this kind ofconditional; the present subsection points ahead to that discussion.
The linguistic description of even if amounts to the intersection of thedescription of if and that of even.63 Consequently, even if has to be accountedfor in the scalar terms which we have by now become familiar with: the presenceof even implicates that the circumstances specified by the conditional are theleast likely for the state of affairs of the main clause to obtain. Just as in the
case of 'simple' even, the illocution (see 2.3.1.2) of an utterance containing evenif is the assertion of a superlative fact. The alternative for the focus constituent,and, hence, a lower value on the scale in question may often be conceived ofas the negation of the sentence functioning as focus constituent. Consider:
(22) He will come even if it is raining.Focus: if it is raining
Scope: He will come ifX
Scale: if it is braining
if it is not raining
But other kinds of alternative are possible as well, especially when one ofthe constituents of the conditional has by itself scalar force, as in He will comeeven if there is a snowstorm. On account of their paradigmatic (scalar) relationwith alternatives, even if -conditionals, unlike normal conditionals, always implythe truth of their main clauses. The superlative nature of that 'truth' is indeedthe very reason for the uttering of the even //-clause. The state of affairs referredto in the main clause may be characterized as superlative ('remarkable') because it is normally ill compatible with the circumstances specified by the con-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 59/318
THE LINGUISTIC BACKGROUND: SCALARITY 49
ditional. In other words, in the case of even if the following statement holds (seeKönig 1986: 233):
(23) In a situation in which it is relevant to say 'Even if p, then q', it isequally relevant to say 'If p, then normally not-q'.64
On account of this reduced compatibility between the main clause and the subordinate clause, the even if -clause may be rightly called the concessive conditional. See further ch. 6.
The notion of 'concession' is one of the central concepts in the study ofscalarity. Concession is always present, whether actual or potential, in scalarexpressions {even or scalar superlatives), because there is always a conflict,overt or covert, between a scalar expression and its scope. In the course of the
present study it will appear that concession plays a major role in the description of per in Homer; it will come into the picture at various points (see 3.2.2;4.3; ch.6). But to speak about per (or any other item) in terms of concession(see 1.3.2) is not very useful as long as the scalar roots of the concept, bothsynchronic and diachronic, are not taken into account. It is with these roots thatthe present chapter is concerned.
2.3.2 Types of focus context
We may say that focus particles create a focus context. Their focus constituent always attracts the main stress of the sentence. It has this property incommon with scalar superlatives. Scalar superlatives, too, whether grammatical or 'pragmatic', typically alter the intonation contour of the sentence in whichthey occur. But focus particles are not the only expressions in which a constituent 'structurally' carries emphasis: the class of focus contexts comprisessome more expression-types, notably comparatives and what may be called 'thesame ay-contexts'. In the present subsection I briefly discuss these two expression-types, with special attention to the semantic properties they have in common with focus particles.
2.3.2.1 Comparatives
The comparative may be described as equivalent to a two-place predicate;65 their second (object-) argument attracts main stress:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 60/318
50 LINGUISTICS AND FORMULAS IN HOMERThe stress on 'Anatoly' is explained by the fact that the second argument of thecomparative may be conceived of as an elliptic clause, whose subject is contrasted (and indeed compared) with the first (subject-) argument of the com
parative.67 'Anatoly' in (24) is the only element by which the 'comparativeclause' differs from the 'main clause':
(24') Garry is a chess player.Anatoly is a chess player.
The element by which the two clauses differ, in the present example the sub ject term of the comparative clause, is highlighted ('emphasized'), for obviouscognitive and perceptual reasons: what is different is perceptually more salientthan what is identical, the different element standing out as a figure against the(background formed by the non-differing material.68 Another, closely related,
property of the comparative (clause) is that it does not admit negation'. '*John plays better chess than Peter does not' is nonsense. Comparison and degree presuppose positive properties. The absence of aproperty has no degree;69 it isabsolute, and perceptually entirely non-salient.70
The behavior of comparatives with regard to negation is characteristic offocus contexts in general. Focus particles do not (normally) allow negationeither. In 2.41 will discuss the topic of negation for its own sake, in connectionwith scalar expressions. And in 3.31 will discuss the comparative in connectionwith scalarity.
23.2.2 The same as and related expressions
The second type of focus context discussed here can be exemplified onthe basis of the following expression-type:
(25) Peter has the same car as JOHN.The expression the same as has the same properties as the comparative: its com plement is the subject (or object) of an elliptic clause which is contrasted to the parallel term in the 'higher' clause. The 'argument' of the same as is highlighted,and is as such a focus context. Here, too, negation cannot occur.
Expressions containing the same as do not stand alone; they belong to alarger group of expressions which have in common that a main clause is identical to a subclause except for one element, for example:
(26) I said to Peter what JOHN had said to Peter.
(27) If you can beat Kasparov, you can beat Karpov.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 61/318
THE LINGUISTIC BACKGROUND: SCALARITY 51
Expressions as these are highly relevant in connection with the study offocus particles. The focus context created by them may be naturally conceivedof as the focus constituent of the inclusive, non-scalar focus particle also or too.Consider, for example:
(26)' I said to Peter what JOHN had said to Peter too.
Here 'John', which is a focus term anyway, serves as the focus constituent oftoo. The only alternative value for the focus constituent of too has already been
expressed as the subject of the main clause (T), that is, the term to which thefocus constituent is contrasted. This is a peculiar use of also/too, which is confined to subordinate clauses. Normally, also/too has the simple existential im-
plicature, which is expressed in (15)B above. For example, in
(28) John has said this to Peter too,
the implicature of too is that there are other people in addition to John whosaid this particular thing to Peter. (Of course there is no scale in the sense thatJohn was the least Ukely person to say this thing to Peter.) In (26) and (26)', on
the other hand, the inclusive nature of the focus term ('Peter') is not so muchimplicated by too ; rather it springs from the nature of the expression as a whole,too in (26)' being after all optional and redundant.71
2.3.2.3 Focus contexts and the study of per
The discussion of the same as and related expressions as focus contexts,and, hence, the treatment of these expressions as a semantic phenomenonwhich is cognate to even and scalarity, appears to be higly relevant for the studyof the particle per in its totality. It provides the basis for the unifying descrip
tion of the particle. Per is the most salient item in the Greek equivalents of exx.(25)-(27) above, that is, in the the same as-type of focus context discussed inthe previous section. Per here is used as an enclitic adhering to the subordinateof the (sub)clause in which the focus term occurs.
This use of per is prominent in Classical (Attic) Greek, but it occurs asearly as Homer. It will not be discussed in the present study. In the presentstudy we are exclusively concerned with what was called in 1.1 'Homeric'per;
post-Homeric (Attic) per we reserve for another study (Bakker: in prep.) Thestudy of 'Homeric' per involves the scalar conceptual framework presented inthe present chapter. 'Attic'per, on the other hand, has to be described as a non-
scalar inclusive focus particle whose distribution is confined to subordinateclauses.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 62/318
52 LINGUISTICS AND FORMULAS IN HOMER
Scalarity is the specific difference between Homeric and Atticper. But thenotion of focus context is their common denominator. Before we can start thedescription of per as a scalar (focus) particle in Greek epic, we have to discussfirst one last aspect of scalar expressions and focus particles alike. This involves the notion of polarity.
2.4 Negative and positive polarity
In the fourth and last section of this chapter I discuss scalar expressions
and focus particles from the point of view of negation. We saw that the focuscontexts discussed in 2.3.2 did not allow negation. In the case of scalar expressions and focus particles, negation does not come as a matter of course either.When negation is discussed in connection with scalarity and scalar expressions,the notions of polarity and polarity item have to be introduced.
Polarity items are lexical items or idioms that are distributionally confinedto polar contexts. Polar contexts are environments that are either commanded
by a negation or, conversely, do not admit negation; accordingly, we divide the polarity items into negative' or 'positive': negative polarity items (NPFs) and
positive polarity items (PPFs). NPFs are distributionally confined to negativeenvironments; they may be, diachronically, characterized as 'strengtheners ofthe negation (Horn 1978a: 148-9). PPFs, on the other hand, are normally confined to affirmative (positive) environments. The difference between negativeand positive contexts is somewhat more complicated than the simple presenceor absence of negatives or negative elements, as we shall see. In the two following subsections I will discuss positive polarity items and negative polarityitems, subsequently.
2.4.1 Positive polarity items
Positive polarity items include a wide range of lexically and categoricallydisparate elements, for example: already, still, relatively, surely, each, swarmwith, some. All these items normally refuse to stand inside the scope of negation, for example:
(29) The Trojans have already won the battle.
This sentence does not readily allow negation. The only way for negation tooccur at all in a sentence like (29) is as an outright denial (see Seuren(1985:233)), in a situation where (29) has been uttered just before:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 63/318
THE LINGUISTIC BACKGROUND: SCALARITY 53
(30) The Trojans have NOT already won the battle (i.e. Your saying "TheTrojans have already won the battle" is unwarranted).
This is the so-called 'radical' or 'external' negation, which takes a sentence asa whole in its scope. It has to be distinguished from 'normal' or 'internal' negation, which is sentence-internal. External negation plays only a very marginalrole in language.
Now, focus particles belong to the class of the PPI's; they refuse to standinside the scope of normal negation. It is true that this contention seems to beimmediately refuted by examples like the following:
(31) A. Only PAUL didn't like the excursion.B. PAUL didn't like the excursion either. Not even PAUL liked the excursion.
But the exception is only apparent. The negation does not have the particle inits scope. It is the other way around: the negation belongs to the scope sentence(see 2.3.1.3) of the particle, which is in all the three cases:
(32) X did not like the excursion.In the farther discussion of polarity phenomena we will dismiss the class offocus particles as a whole to deal exclusively with even.
From the examples in (31) it is clear that the difference between positiveand negative polar contexts is not simply the presence or absence of negation.In (31) we have, in spite of the presence of a negation, a positive polar context,created by the focus particle. The same applies to sentences containing scalarsuperlatives. Here, too, a negation may very well occur:
(33) The most complex problem doesn't pose difficulties for him.
The negation here simply is not 'negative', the example being equivalent to Hecan solve the most complex problem (see ex. (3) in 2.1.2 above).
The analysis of even and other scalar expressions in terms of positive polarity will prove to be relevant later on for the discussion of per and kai inconnection with negation (see 3.2; 4.2.2; 4.3). For the moment, however, theinsight that negation may occur in contexts of the PPI even (viz. in its scope) isimportant in connection with the notion oF scale reversal. In the work of Fauconnier on scalarity it is repeatedly stated that when we compare sentences like
(34) The simplest problem poses difficulties for him
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 64/318
54 LINGUISTICS AND FORMULAS IN HOMER
to (33) above (see also (3)-(4) in 2.1.2), we have to say that the negationin the one sentence reverses the scale of the other. Thus the scale associatedwith (34), with the simplest problem as its high point would appear in (33) inreversed form, with the most complex problem as highest value. It is true thatnegation may have, from a logical point of view, a truth-value reversing effect(the negative sentence being false when its affirmative counterpart is true, andvice versa), but Fauconnier's contention seems to me basically incorrect. In(33) and (34) we do not have one and the same scale which is reversed accord
ing to the presence or absence of negation. Rather, we have to say that thereare two different scales, associated with two different scope sentences, the onecontaining a negation and the other not. The scale associated with the negatedsentence is the scale of the complexity of problems, while the one associatedwith the affirmative sentence is the scale of the simplicity of problems. Asregards the nature of the scale, there is no difference, and scale reversal doesnot occur. Both scales are positive in nature and have only one high point. The
presence of negation in the scope sentence of one of them is immaterial.
The question as to whether a scale is the reversed form of another mayseem mere hairsplitting at first sight. However, this objection disappears whenwe turn to negative polarity. Here a phenomenon occurs which may rightly becalled scale reversal, and it is quite different from the alleged reversal in (33)-(34).
2.4.2 Negative polarity items and scale reversal
The particle even in English differs from its Dutch and Germanequivalents zelfs and sogar among other things in that an account purely in termsof positive polarity does not exhaust its possibilities: even may also occur innegative contexts (under the scope of negation), unlike zelfs and sogar. But ifit does, a curious phenomenon may be observed, which may be called scalereversal. This phenomenon should be discussed in connection with negative
polarity and negative polarity items (NPFs), and it is closely related with thenotion of wide scope of the particle even (see 2.3.1.5).80
NPFs require a negative or a negative context, on pain of ungram-maticality. NPFs in English include ever, at all, any, anymore, in weeks, in theleast, mind that etc. These and other NPFs may occur in contexts commanded by an overt negative, but also in contexts with a covert or implied one. The latter include questions (especially rhetorical ones), conditionals {If anyone..),
and elements which expHcitly exclude something, like superlatives ('that's the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 65/318
THE LINGUISTIC BACKGROUND: SCALARITY 55
greatest one I ever saw'), only, first, until (punctual), as well as inherently negative predicates like refuse, prevent etc.
Of particular interest for my present purpose are those expressions amongthe NPI's which denote a minimal quantity or notion: the slightest, (not) a bit,soul etc., lift a finger, give a damn etc. These expressions cannot occur withoutnegation; they serve indeed as negation strengtheners, often enhancing theemotional impact of negation (for negation strengthening see 3.4.4 below).
The negative polar expressions in question have to be discussed in con
nection with scalarity and even. It is obvious why this is so: if even the smallest possible portion or quantity is not the case, then any greater portion orquantity will, a fortiori, not be the case either. Thus these expressions are intimately related to superlativeness and scalarity. The NPI the slightest is evenmorphosyntactically a superlative; it may be described as a scalar superlativewhich is distributionally confined to negative polar contexts.
The occurrence of scalarity in negative polar contexts has an importantconsequence for the scale involved. It is here that the notion of scale reversal isin its proper place: the scales associated with scalar expressions in a negative
polar context have no positive extreme value, a high point, but a negative extreme value, a low point, so to speak. A scalar expression in a negative environment represents the extreme, minimal, value on a scale; all other valuesimaginable are higher on the scale, but because the scale is a reversed one, theyare implied by, and hence, less salient than the low point.
Reversed, negative polar scales are marked with regard to 'normal', positive polar ones. This can be gauged from the fact that in the case of negativescalar expressions a lexical constraint obtains which does not exist in the caseof positive scalar expressions: negative scalarity and reversed scales can only
be expressed by terms which denote lexically a mimimum quantity or quality:
beside pragmatically inverse superlatives like lift a finger, bat an eyelid we maymention slightest, tiniest, remotest, faintest. This is a lexically homogeneous class,of which the slightest is the member which is most widely and regularly used asa NPI. In the case of positive polar scalarity, on the other hand, there is no lexical constraint. Here any superlative may occur, regardless of its lexical meaning. Both biggest and smallest may be used as positive scalar superlatives. Eachhas its own scale, and both are the positive high point of it, not simultaneouslythe high point and the low point of one and the same scale. The notion of scalereversal has to be reserved for the qualitative difference between positive polar
and negative polar scales.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 66/318
56 LINGUISTICS AND FORMULAS IN HOMER
We now return to the particle even. English differs from other languagesin that it does not use its scalar focus particle exclusively as a PPL Even mayalso be used in negative polar contexts. But if it does, the mechanism of scalereversal is put into practice: the focus constituent of even represents the low
point of a scale, instead of the high point. The notion of scale reversal should be seen in connection with the fact that in exx. (20)-(21) in 2.3.1.5 even in thewide scope reading implicates the opposite of even in the narrow scope reading: the positive scale of which Homer as the most difficult author is the high
point is reversed into a negative scale of which Homer as the least difficultauthor is the low point. Even in negative contexts is the extensional counter part of elements like the slightest.
German and Dutch differ from English in that there is a separate expression for scalarity in a negative context: auch nur and ook maar. In section2.3.1.5 above, I discussed these particle combinations as wide scope particles.We may now go a step further and analyze them in connection with the notionof scale reversal: the first element (auch/ook) expresses the scalarity (implication), while the addition of nur/maar denotes the minimum quantity orquality, and hence the reversal.
Negative polar scalarity is of less practical importance for the descriptionof per in Homer than positive polar scalarity. There exists, however, a numberof examples to which the above discussion neatly appUes, see 3.4.3 below.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 67/318
NOTES TO CHAPTER 2
1) In this connection it is interesting to point to the fact that theos may be modified by the
suffix -teros (theôteros). This suffix, which later became the normal comparative suffix in Greek,
originally marked binary oppositions and was distributionally restricted, according to Benveniste
([1948] 1975:117) to the marked term of such oppositions. Thus the existence of theôteros (e.g.
Od. 13, 111) but not of *anthrôpoteros would seem to be an index of the marked status of theos
in the opposition. I owe the content of this note to prof. CJ. Ruijgh (p.c.).
2) For a pertinent discussion of the distinction between encoding devices (viz. (mor-
pho)syntactic phenomena) and encoded phenomena (semantic) see Tomlin (1983: 416-7).
Similarly, Traugott (1985: 290 ff.) distinguishes between a semantic, abstract notion of
'conditionality' and the markers encoding it.
3) For elementary exposition on paradigmatic and syntagmatic) relationships see Lyons
(1977: 240-41).
4) See Horn (1972), Gazdar (1979), Levinson (1983: 132 ff.). Within the framework of
'conversational implicature', scales are used as follows. They consist of items which can be ar
ranged by degree of informativeness or semantic strength, for example all, some. A speaker using
a weaker (rightwards) item on such a scale conversationally implicates that more stronger
(leftwards) items do not apply. And his addressee, assuming co-operation on the part of thespeaker, infers that the stronger item does indeed not obtain, in spite of the fact that the speaker
has not explicitly said so. Thus in this type of scalar theory the highest item but one (or more) is
the thing that matters, whereas in the scalar theory used in the present study we are exclusively
concerned with the highest item.
5) See Van der Auwera (1983,1985), Ducrot (1980), Horn (1972). The study of particles
in this connection is mostly concerned with German (Altmann (1976), König (1981),
Jacobs(1983)) and Dutch (see for example Van der Auwera & Vandeweghe, eds. (1984)). For
English see Taglicht (1984).
6) Notice that this account of scales like (2) agrees well with the marked status of 'god' in
the binary opposition with 'man' (see fn. 1).
7) See Fauconnier (1975a, 1975b, 1979). For particles ( = even) in this connection seeKarttunen & Peters (1979) and König (1981). See 2.3.1 for more literature on even.
8) 'State of affairs' I use in the present study as that to which a proposition refers. A state
of affairs consists of a certain relation between a number of entities. The nature of this relation
is expressed by the predicate of the sentence by which the proposition is uttered. The arguments
of the predicate are the terms referring to the items in the state of affairs in question. Thus the
proposition (or predication) expressed by John gave the book to Peter refers to a particular state
of affairs consisting in the event characterized by 'John's' giving a particular book to 'Peter'. The
predicate 'give' is three-place because it obligatorily has three arguments. The notions just given
can be found in a number of linguistic schools and theories which have in common their treat
ment of the predicate as the central part of the sentence. See for instance Dik's (1978: 25-31)
notion of 'nuclear predication' (= predicate + argument(s)). For a general survey see Matthews (1981:121 ff.).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 68/318
58 LINGUISTICS AND
FORMULAS IN HOMER9) Notice that a literal reading of (3)-(4) is easier when we take 'problem' in the plural:
He couldn't solve the simplest problems might then well mean that the person in question was not
able to solve, on a particular occasion, problems that could be characterized as Very simple'.
But this reading does not exclude his actually having solved a very difficult one.
10) Notice that 'the simplest problem' and 'the most complex problem' are the high point
of two different scales. At first sight one would assign the two superlatives to the extremes of one
and the same scale, ranging from 'difficult' to 'simple'. This is, however, impossible. As has been
said, scales must be unequivocal in orientation. A scale which has at the same time two potential
high points cannot exist in the scalar framework presented here. See Horn (1972:39) for the im
possibility to rank hot/warm and cool/cold on one and the same scale. Scales of the type under
study do not tolerate the conflation of two dimensions/orientations. See further 2.4 and 3.4.3
below.
11) See Fauconnier (1975a) for a detailed account of this similarity. Fauconnier shows
that the distribution of scalar superlatives is that of quantifiers, and not of normal definite
descriptions.
12) See Horn (1972), Davison (1980).
13) Notice that when meant as a scalar statement, (5)-(6) as utterances have an entirely
different intonation pattern from the literal meaning: 'Kasparov' and 'Getty' attract the main
stress of the sentence.
14) What classicists commonly call 'irrealis' is in logic and general linguistics usually
referred to as 'counterfactual' (conditional).15) In a sentence like The bravest man would have been afraid (on that occasion) there is
specific temporal reference in an indirect way, because a particular situation is meant. Yet the
sentence has to be ranked among the irrealis phenomena, since there is no question of a specific
state of affairs in which a particular brave man was actually afraid.
16) No item of the triad 'tense-aspect-modality' can be fruitfully discussed and studied in
isolation from the other two. See the studies in Hopper, ed. (1982), especially Wallace (1982),
as well as Givón (1984: ch. 8).
17) This is not to say that future statements cannot possibly be factual. In fact they often
are. But it is never by virtue of the future tense marking that they are factual. The factuality of
future sentences is due to other factors, such as when a future predicate is embedded under a
factive predicate (in the sense of Kiparsky & Kiparsky 1968) like 'know*. Sometimes future statements are factual on the level of the intention of the speaker uttering them (this involves the no
tion of illocution, see 2.3.1.2 below): when a future occurs in an appointment ('I'll be there at
eight'), or a threat or a warning, its factuality is bound up with the felicity of the speech-act in
question, to the effect that when there is doubt as to the factuality of the utterance, the speech
act may be considered unsuccessful. But by their very meaning, futures belong to the realm of
irrealis. See the discussions in Fleischman (1982).
18) In the case of meaning differences along the axis 'certain-uncertain' we may speak of
epistemic modality (see for instance Lyons (1977:793 ff.), Palmer (1986: ch. 2)). A typically un
certainty mood (and hence a case of irrealis modality) is the Greek optative. For a discussion of
epistemic modality as a typical main clause phenomenon see 4.1 below).
19) A state of affairs that has to obtain does not, by definition, obtain already. There is a
strong semantic and ontological connection between deontic modals and the future. Sometimes
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 69/318
NOTES TO CHAPTER 2 59
future tense paradigms evolve diachronically from deontic modals, as is the case in Romance(see Fleischman 1982).
20) These verbs are called non-implicative modality verbs' (Givón 1973a: 98-101; 1973b,1984:117-9): commitment to the truth of the predicate itself does not imply commitment to thetruth of the clause embedded under it. The truth of 'Peter wanted to leave' does not entail thatPeter actually left. The complemeent of implicative modality verbs, on the other hand, is facially modalized: the truth of' Peter managed to arrive in time' does imply that Peter actually arrivedin time. Notice that just as in the case of deontic modals (see the previous note) there is an in
timate bond between non-implicative modality verbs and the future. In some languages (e.g.Modern Greek) the future evolves diachronically from 'want'. In Ancient Greek, too, the futureis often conceived of as the offspring of an earlier 'desiderative' mood. See in general Fleischman(1982).
21) See Givón (1973:109-110; 1984: 285).
22) Here non-fact modality appears as the crucial common property of future and generic present sentences. In Greek the subjunctive conditional (ean 4- subj.) is distributionallyrestricted to main clauses with future or generic present predicates.
23) See in detail Givón (1973a; 1984:285,391 ff.).
24) Sometimes (e.g. Hawkins 1978) the term 'specific' is used to designate referential noun
phrases whose referent is not assumed to be known to both participants in a conversation.25) As a parallel phenomenon beside the connection between realis modality and referen-
tiality we may mention the connection between telic aspect of the verb and the individuation ofthe object: verbs denoting actions with an inherent end-point tend to have more fully individuated and referential objects than verbs with atelic aspect. Both realis modality and telicaspect may be seen as an index of the clausal property of 'transitivity', the degree to which anaction is 'carried over' from an agens to a patiens. For transitivity as a property (allowing of degree) of clauses in discourse, see the seminal discussion in Hopper & Thompson (1980).
26) Polarity contexts are contexts in which 'negative polarity items' may occur. Existentialany belongs to these, just as ever, at all and predicates like 'give a damn'.
27) Notice that indefinite noun phrases in negated and 'hypothesized' sentences tend tohave a non-referential understanding. See Givón (1973a; 1979:93 ff.).
28) But not the deontic modals must and have to in commands and adhortations (i.e. indirective speech acts, see 3.2.2): *'Anyone must leave'.
29) The only way for universal any to occur in factual contexts (progressive present, pasttenses) is when it is modified by a restrictive relative clause: *'He bought anything' is unacceptable, but 'He bought anything he could get' is not. In other words, any in irrealis contexts maystand on its own, whereas in realis contexts it needs restriction. For more details on the distribution of any see Horn (1972:131 ff.) and Davison (1980:11-15).
30) We may say that it is the prototypical adjective that encodes properties, whereas it isthe prototypical verb that encodes actions and the prototypical noun that encodes permanent
states. The notion of prototypicality implies that adjectives may as to their meaning be more'noun-like' (i.e. to a higher degree denoting a 'state') as well as more Verb-like' than the prototype, depending on the 'time-stability' of the entity that they denote. See further 4.3.2 below.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 70/318
60 LINGUISTICS AND FORMULAS IN HOMER
31) It should be noted, however, that 'Peter's best friend' in (11) as it stands allows of a
referential understanding as well. In that case, (11) may be read as an equative sentence, in which
the identity of two entities is asserted, just as in mathematical equations (see Lyons 1977:185,
471-2). In Bakker (1986; in prep.) I make the distinction between equative and ascriptive sen
tences productive for the study of per in Attic Greek.
32) See Strawson (1950: 320-1), Givón (1984: 91,396).
33) It should be mentioned at the outset of this subsection that in discussing such topics
as reference, intension and extension my purpose is purely expository. Consequently, the dis
cussion is, and is meant to be, informal. The very numerous philosophical controversies and sub
tleties connected with reference, proper names, de dicto and de re interpretations of definite
descriptions etc. are, on the whole, irrelevant for my purpose. They are indeed irrelevant for
many linguistic purposes when one is concerned with concrete facts of language.
34) Notice that philosophers are, on the whole, impervious to notions like relevance, ap
propriateness and felicity, in short, the pragmatic component of reference. They tend to stick to
the truth-conditional substitution principle. A case in point is Donellan ([1966] 1971), who
rigorously distinguishes between a referential use of definite noun phrases on the one hand, and
an attributive use on the other: "In the referential use the definite description is merely one tool
for doing a certain job - calling attention to a person or thing - and in general any other device
for doing the same job, another description or name would do as well. In the attributive use, the
attribute of being the so-and-so is all important, while it is not in the referential use" ([1966]
1971: 102). Some things are kept apart here which had better be taken together. For more on
these matters see Schoorl (1980), Bakker (in prep. ch. 3). I am aware that any critique of the
rigidity inherent in the philosophical approaches to reference is stating the obvious in the eyes
of those linguists who are concerned with reference and definiteness in the wider context of com
munication and discourse. But as the terminological apparatus I am using is common to both
philosophers and linguists, some elementary points have to be made.
35) It is perhaps superfluous to state that this person or thing need not actually exist in the
real world. Reference in language is intimately connected with the intentions of speakers; as
such it involves a universe of discourse, rather than the 'real world'. See Givón (1981; 1984:388-
90).
36) Notice that I conceive of 'reference' as an activity of speakers. Extension, on the other
hand, I treat treat as an inherent property of noun phrases. See further below.
37) Normal English usage: 'sense' (intension) and 'reference' (extension). In Frege (1892)
'Sinn' and 'Bedeutung'. Also used are 'connotation' and 'denotation' (first in Mill 1843). See in
general Lyons (1977: ch. 7) and Haack (1978:246). Already in Antiquity, the Stoics were aware
of something in between the name itself (semainon) and the thing referred to (semainomenon);
they called it lekton (see Kneale & Kneale (962:140)).
38) Notably in the form of the distinction between 'extensional' and 'intensional' contexts.
Extensional contexts are environments in which substitution of co-extensive terms has no truth-
conditional consequences (it does have appropriateness-conditional consequences, but that is
irrelevant in logic). In intensional contexts, substitution of co-extensive terms is impossible
without changing the truth-conditions. Acknowledged examples are contexts under the scope
of modal operators ('necessarily....') and the complement of predicates expressing a thought
('think', 'know', 'believe': 'John knows that Bill died last week' does not have the same 'truth-
conditions' as 'John knows that Peter's best friend died last week'. The direct object of a predi-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 71/318
NOTES TO CHAPTER 2 61
cate like 'kill', on the other hand, creates an extensional context. See Frege (1892), Quine (1953).Critique and reassessment in Seuren (1985: ch.5).
39) A non-exhaustive Hst: Horn (1969), Fraser (1971), Fauconnier (1975a, 1975b),Kempson (1975:200-2), Karttunen & Peters (1979), König (1981), Bennett (1982), Jacobs (1983:144 ff.), Seuren (1985:310-313).
40) As often, a different view is adopted in Seuren (1985). Seuren argues that the meaning of even and also/too is truth-conditional.
41) In later publications Givón discusses these discourse-presuppositional phenomena interms of topic continuity, see Givón, ed. (1983).
42) For example, 'the x' presupposes that ' ' actually exists ('existential presupposition').Another 'presupposition-trigger' is the group of factive verbs (see Kiparsky & Kiparsky 1968):'He regretted drinking much' presupposes that he drunk much. This kind of presupposition(often called semantic presupposition) is characterized by the so-called 'constancy undernegation': the negation of the sentence in which they occur does not affect them. For elementary exposition see Levinson (1983: ch. 4). See also Horn (1972), Kempson (1975).
43) Much research in focus particles and polarity phenomena (see 2.4 below) is essentially semantic, non-functional and formal. Karttunen & Peters' discussion of even and other casesof 'conventional implicature', for example, is accompanied by an unnecessary formal(istic)analysis within the framework of Montague Grammar. Yet in spite of Givón's (1982:85) verdict,I believe that Karttunen & Peters' discussion has its merits. Especially the concepts of 'focus'
and 'scope' are valuable in the study of focus particles.44) Grice (1975) is a partial publication of Grice's William James Lectures, delivered at
Harvard University in 1967.
45) The cooperative principle can be subdivided into four categories, Quantity, Quality,Relation and Manner (Grice 1975:45). The four categories are related with certain 'maxims ofconversation'. A simple example, taken from Grice (1975: 51): a person who says to someonewho is out of petrol 'There is a garage round the corner' conversationally implicates that thegarage in question is open (at least in his opinion) and actually has petrol to sell. The personwho is in need of petrol normally will understand the utterance as such, assuming that his conversation partner conforms to the maxim of relevance, as it is a wholly uncooperative and irrelevant thing to mention a garage which one knows is closed at the moment. Notice, however,that conversational implicatures are always so loosely attached to an utterance as to be easilycancellable. There is no inconsistency when one says, in the above situation, 'There is a garageround the corner, but that one is closed today'. See further, apart from Grice (1975), Karttunen& Peters (1979:2, n.3), Levinson (1983: ch. 3).
46) Notice, thus, that conventional implicatures are not cancellable in the same way asconversational implicatures (see the previous note). A speaker denying (15)B-C after uttering(14) is self-contradictory.
47) Except when they are used inferentially, viz. when an inference is drawn from a proposition whose truth has already been ascertained earlier in the discourse or conversation ('A is thecase'. 'Oh, if A is the case, then is (also) the case'). See Bakker (1988a: n. 21), and in more
detail Bakker (in prep. ch. 7).48) See Baker & Hacker (1984: ch. 5) for severe criticism of truth-conditional semantics.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 72/318
62 LINGUISTICS AND FORMULAS IN HOMER49) In this connection the use of the term 'pragmatic' may give rise to terminological con
fusion, as in certain linguistic circles it is customary to use the term 'pragmatics' for all non-truth-
conditional aspects of the meaning of sentences (including implicatures) and 'semantics' for
truth-conditional phenomena. I prefer to use the term 'pragmatic' for the use that is made of
sentences in speech situations and in coherent discourse (see Givón 1984:30-1). For a general
discussion of the delimitation of 'pragmatics' with regard to 'semantics' see Levinson 1983:5-35.
50) Notice, in this connection, that even and scalar superlatives can co-occur, the scalar
superlative being the 'focus constituent' (see 2.3.1.3) of even: 'He could solve even the most dif-
ficult problem'. In that case even is, strictly speaking, redundant, and hence a peripheral instance
of its kind (see 1.4); the combination of even with a scalar superlative has to be discussed in termsof 'strengthening'.
51) See Lyons (1977:725 ff.); Bolkestein (1980:31).
52) See Karttunen & Peters (1979:15).
53) This is not to say, of course, that other constituents of the sentence cannot or do not
carry emphasis.
54) 'Focus constituent' is a semantic notion. The syntactic entity associated with particles
like even is called in König (1981:109) 'sister constituent'. Normally, the focus constituent will
coincide with the sister constituent. Sometimes, however, there is a difference. biEven his easiest
books are difficult to understand the sister constituent of even is 'his easiest books'; the focus con
stituent, attracting main stress, is 'easiest'. Sometimes (e.g. Bennett 1982) what is called here'focus' (constituent) is referred to as 'scope'. This is an unusual practice.
55) This is not to say, however, that zelfs is unacceptable as a translation of even in (19).
My point is merely that Dutch has a choice here, and English not.
56) The whole scalar 'correlation' is even.let alone; schon...geschweige denn (German);
alleen allnog.laat staan (Dutch). Thus we see that complex expressions, the meaning of whose
constitutive parts has no independent importance, may function as genuine scalar particles with
their own focus constituent.
57) See section 3.3 below for yet another case in which English has to use even where other
languages have another particle at their disposal: the second argument of a comparative predi
cate as the focus constituent of even ('even taller than x '.
58) See Karttunen & Peters (1979:25), König (1981:122), Fauconnier (1975b: 190).
59) Notice that this type of ambiguity is not confined to scope sentences of even; it occurs
also when a main clause to which a causal subclause is attached is negated, for example: 'I will
not go to the party(,) because I feel sick'. See also the ambiguity in (21).
60) This is what I call elsewhere (Bakker 1988a) a 'non-embedded conditional. In the case
of non-embedded conditionals, two propositions which are in themselves complete stand in an
implicative relationship to one another: 'if proposition ρ is true, then proposition q is true', or
'p implies q'.
61) The determination of the circumstances under which a given state of affairs obtains is
a phenomenon which is typical of irrealis modality and the absence of specific temporal
reference (see 2.1.4). In
the case of realis modality this determination is
unnecessary, on accountof the specific temporal reference of the predicate. Conditionals with a restrictive function may
be called 'embedded' syntactically (see Bakker 1988a). To my mind embedded conditionals have
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 73/318
NOTES TO CHAPTER 2 63no propositional characteristics of their own. The embedded conditional appears in Greek as
'ean + subjunctive'. My distinction between embedded and non-embedded conditionals is
reminiscent of Van der Auwera's (1983, 1985: 209-10 distinction between 'contingency
conditionals' and 'indeterminacy conditionals.'
62) See Haiman (1974: 351 ff.); König (1986).
63) Accounts which try to describe even if as an idiomatic collocation run into difficulties.
For an account in terms of if and even separately see Bennett (1982).
64) See also Thompson & Longacre (1985:197).
65) Sometimes comparatives are one-place on first sight, but this is contextually condi
tioned: utterances like 'John is taller' can be used only when the length of someone else is the
topic of discourse or conversation. See ex. (26) in 3.3 above.
66) Unless, of course, there is a contextually conditioned stress on other constituents of
the sentence.
67) See extensively Quirk et al. (1972:765-77). For 'ellipsis' in comparative clauses: 1972:
768-9. The comparative construction has attracted a great deal of attention in generative syntax
and formal semantics. See Von Stechow (1984) and the various reactions to it in vol. 3 of the
Journal of Semantics. In Hoeksema (1983; 1984), it is argued that what are considered in the text
to be two sides of one and the same phenomenon (viz. the second argument of the comparative
as subject of the elliptic comparative clause) are actually two different comparative construc
tions: 'Bill is taller than Mary' vs. 'He is bigger than we are'. This counterintuitive idea (no functional explanation is offered) has met with little consent, as far as I can see, even in the kind of
linguistic circles where one is not interested in functional explanations.
68) A whole body of literature dealing with the linguistic implications of this insight, and
based on the distinction from cognitive psychology between 'figure' and 'ground', begins to
develop. For example Hopper (1979), Wallace (1982) (where psychological references are
provided), and especially the work of Givón.
69) For more about the positive nature of comparativeness and superlativeness see ex.
(25) with discussion in 4.3.2.1 below.
70) See the illuminating discussions of Givón on the 'ontology' of negation (1979: ch. 3;
1984: ch. 9).
71) For more discussion about the semantics and pragmatics of the same as and the other
expressions in question as well as of the distribution of too and also in these contexts see Bak
ker (1986; in prep.).
72) The difference between the Homeric and the Attic usage is that in Homer per (often)
adheres directly to the focus term (e.g. I say what YOU (per) say), while in Attic the particle al
ways adheres to the subordinator (I say what (per) YOU say, I say the same that (per) YOU say).
See exx. (15)-(16) in 1.2 above for an illustration of the two possible places of per in this expres
sion-type.
73) In other words, Greek has a separate particle to express the use of too in (26)'. But
per also occurs in sentences like (25), where English does not use a focus particle (see ex. (16)
in 1.2 above).
74) More than any other subject touched upon in this chapter, the notion of polarity is the
domain of formal semanticists. Especially the distinction between 'downward entailing' and 'up-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 74/318
64 LINGUISTICS AND FORMULAS IN HOMER
ward entailing' contexts (not to be further discussed here, see Ladusaw (1979), Hoeksema (1983)and Zwarts (1984)), which is adduced in this field as the explanation of the distribution of negative polarity items, seems to be entirely logic-bound and dysfunctional.
75) See Horn (1978a: 154) and especially Seuren (1985: 233).
76) The notion of external negation is most at home in the works of philosophers and formally oriented linguists. It has played a role in the discussion about the presuppositions ofdefinite descriptions. See Russell (1905) and Strawson (1950). More exposition and discussionin Kempson (1975: 95-100; 1977:145 ff.), Levinson (1983:171-72), Seuren (1985: 238).
77) For a discussion of sogar in terms of positive polarity see Jacobs (1983: 211 ff.).78) Notice that not may be treated, paradoxically, as a PPI (see Seuren 1985:233,261).
79) Fauconnier (1975a: 362; 1975b; 1979: 290-91).
80) We may say that the capability of even to occur both in negative polar and in positive polar contexts is closely connected with its being ambiguous as to its scope. Sogar and Zelfs, bycontrast , seem to allow only of narrow scope readings (see 2.3.1.5). Hence they can only be associated with positive polarity.
81) See Seuren (1985:232). NPI's form a just as disparate group as the PPI 's. They includeadverbs, predicates, and quantifiers.
82) See Horn (1978a: 151-52). The application of the notion 'negative environment' is not
a sharp-cut one, and the question as to whether a given NPI is used 'correct!/ or not is often amatter of subjective judgment.
83) See Horn (1978a: 149), Fauconnier (1975:196), Schmerling (1971).
84) Fauconnier always treats the extreme value of scales in terms of 'low point'. I preferito speak in terms of 'high point' and to reserve the term 'low point' for negative polar scalarity,so as to bring out the difference between the two 'poles'.
85) Negatives are marked with respect to the corresponding affirmatives in many ways.For example, from a pragmatic point of view, they presuppose more than affirmatives: a sentencelike Givón's (1979: 103) My wife isn't pregnant presupposes more situational backgroundknowledge than its corresponding affirmative version: in being negative it removes information.But for information to be removed it must simply be there. Positive information may be moreeasily introduced without implicit assumptions. Furthermore, it is easy to see that negatives havealso morphologically a marked status.
86) For some discussion of the distribution of ook maar, see Hoeksema (1983:406 ff.) andPaardekoper (1979).
87) On their own nur and maar are exclusive focus particles; like only they exclude alternatives for their focus constituent. And wherever something is excluded, there is an implicit negation. For a discussion of nur (and erst) in German see König (1981:119) and of {alleen) maar inDutch Van der Auwera (1984).
88) The use of even seems to be similar to that of any. Usually two kinds of any are recognized, an existential and a universal kind (see 2.1.3 above). The latter is sometimes called 'free-
choice any'- Sometimes every connection between the two any's is denied, with the result thatany is thought to be ambiguous (or polysemous), see Horn (1972), Carlson (1981)). Or an analysisis proposed in which any is treated as either of the two logical quantifiers, existential or universal (see Davison (1980) for such an analysis and for discussion of others). Now 'existential' any
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 75/318
NOTES TO CHAPTER 2 65
exclusively occurs in negative contexts and is thus a NPI (If anyone..., Does anyone...? etc.).'Universal' any, on the other hand, seems to be a PPI, a fact which is not often recognized. Onthe basis of polarity and the kind of symmetry that is implies, it seems to me possible to describeany in a unified way, in a manner which is parallel to the notion of scale reversal discussed here.To do this, however, falls outside the scope of the present study.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 76/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 77/318
3 SCALARITY AND THE PARTICLE PER
3.0 Introduction
The present chapter contains the appHcation of the theoretical apparatus presented in the previous chapter. Examples will be discussed in which per hasto be described as a scalar particle. The distinctions between intension and extension (see 2.2.2) and between scope dependence and scope independence(see 2.3.1.4), will turn out to be useful tools for the description of the centraluses of per, as well as for the delimitation of the particle with regard to kai
In section3.11 will deal with ex. (1)-(2) of 1.2 above and similar examples.It will appear that the combination, in a non-factual context, (see 2.1.4) of perwith an intensional term (adjective) functions as a scalar superlative. In section3.2, example (3) of 1.2 and the expression-type to which it belongs will be discussed. Here I will deal with what happens when per is used in connection witha scalar relationship between extensional terms, viz. proper names. Section 3.3contains the use of per in scalar comparative expressions. In 3.4 some exampleswill be discussed to which the concepts of wide scope, scale reversal and negative polarity (see 2.3.1.5 and 2.4.2) apply. Something which resembles scale
reversal in non-factual contexts will play an important role in section 3.5 also.Here the group of instances of which ex. (4) and (5) of 1.2 above are memberswill be discussed. This last use is the only one whose discussion has not been prepared in the previous chapter. This is because the semantics of even cannot be straightforwardly applied to it. In 3.6, finally, some less easily classifiable instances will be presented.
The scalar instances of per to be discussed in the present chapter represent only a relatively limited part of the total material involved. For example,out of a total of more than 500 instances of per in the Iliad and the Odyssey, only
8 are of the type represented by ex. (1)-(2) of 1.2 (see the Appendix). Yet inspite of their small numbers, the scalar instances occupy a central place in the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 78/318
68 LINGUISTICS AND FORMULAS IN HOMERdescription because the much more frequent concessive uses of the particle,especially the expression-type 'participle + per'. are intimately related to them,
both from a synchronic and from a diachronic point of view.
3.1 Intensional terms
3.1.1 Scalar superlatives
Consider the following examples. Per here accompanies an intensional
term, viz. an adjective. The examples conform to a pattern which may be characterized as follows. Per and the intensional term occur in a non-restrictive('digressive') relative clause. The relative clause is modified by the particle te(on which see further 3.1.3 below), which immediately follows the relative (oranaphoric) pronoun.
(1) toi de Phobos philos huios, hama krateros hai atarbês,hespeto, hos t' ephobêse talaphrona per polemistên.(Ares the Killer and his son,) "the fierce and indomitablePanic-maker, before whom the staunchest warrior turns tail." (II.
13,300).(2) enth ' eni men philotês, en d' himeros en d' oaristus
parphasis, hê t' eklepse noon puka per phroneontôn
(The girdle of Aphrodite: it contains) "Love and Desire and thesweet bewitching words that turn a wise man into a fool [Ht: that
steal away the mind of the wisest people]." ( . 14,217).
(3) hai kholos, hos V epheêke poluphrona per khalepênai"Anger, that makes the wisest man flare up." (II. 18,108).
(4) gê phusizoos, hê te krateron per erukei"the fruitful Earth ( = the Netherworld), who holds the strongest
man down." (II. 21, 63).
(5) oinos gar anôgei,êleos, hos t'epheêke poluphrona per mal1
aeisai
"This is the effect of your wine - for wine is a crazy thing. It makesthe wisest man sing." (Od. 14, 464).
What the relative clauses in these examples tell about the extension (see 2.2.2)
of their head noun is, in being something remarkable, a superlative attribute:Phobos' in (1), for example, is extremely fearsome and the effect of wine is
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 79/318
SCALARITY AND THE PARTICLE PER 69
said in (5) to be wholly irresistible. Each of (1)-(5) is the expression of a superlative fact (for this term see 2.3.1.2 above).
In 2.1.2 and 2.1.3 above we saw that scalar superlatives naturally occur insentences functioning to ascribe some superlative attribute to a person or athing (e.g. extreme cleverness in the case of He could solve the most complex
problem). In other words, scalar superlativeness is used in service of the expression of attributive superlativeness. Now it is impossible to escape the conclusion that this is exactly what happens in (1)-(5). The adjectives function as
scalar superlatives in sentences that ascribe some superlative attribute to someone or something.
In English, the scalar function of an adjective is marked by morphosyn-tactic means: the adjective appears in the superlative mode. Consequently,there is in EngUsh no formal difference between scalar superlatives on the onehand, and attributive or referential superlatives on the other. In (Homeric)Greek, on the other hand, there is a different strategy. Here the scalarity of theadjective is marked by the particle per , the adjective itself remaining in the normal positive mode. Per as it functions in (1)-(5) is a curious way to expressscalar superlativeness. In terms of the previous chapter, we may describe it asa scalar particle which takes as its focus constituent an intensional term thatrepresents the high point of a scope-independent scale.
In exx. (1)-(5) extensionality is completely absent, the focus constituentsof per being adjectives, and as such entirely intensional. The following example, on the other hand is different, in spite of its similarity in form, in thatthe term preceding per is a substantive:
(6) smerdale' eurôenta, ta te stugeousi theoi per
(The Netherworld:) "The hateful Chambers of Decay that fill even
the gods with horror." (77. 20, 65)On account of the translation ('even the gods') it might seem on first sight thatwe have here a statement in the format of ex. (14) of chapter 2: Even Paul liked
the excursion. However, to attach too much importance to the translation of (6)would be to obscure two essential differences between (6) and ordinary even-sentences.
First, unlike 'Paul', theoi ('gods') in (6) is not purely extensional: besidean extension, it has intension as well. This appears from the fact that 'god' formsa binary pair with 'man' (see 2.1.1). In this binary opposition 'gods' is the
marked term, which yields the intension ' + immortality' (as opposed to '-immortality' for 'man'). That theos is not a prototypical extensional term can
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 80/318
70 LINGUISTICS AND FORMULAS IN HOMER be gauged from the fact that it may be modified by the suffix -teros, which isotherwise confined to the adjectival word. 'Paul' (or any other proper name),on the other hand, has no such meaning.
The second important difference between (6) and the even-sentence EvenPaul liked the excursion is that the latter is factually modalized, whereas (6) isan instance of non-fact (irrealis) modality (see 2.1.4). Even Paul liked the excursion is a proposition referring to a particular, existing state of affairs. Ex. (6),on the other hand, is generic; it does not refer to any specific state of affairs in
which the gods are or were afraid of Tartarus.It follows from this that (6) belongs to the expression-type exemplified by
(l)-(5); it possesses most of the characteristic properties of the scalar superlative: intensionality, non-factual modality, and scope-independence of the scalein question. It is possible, on account of the not fully intensional character oftheoi, to assign (6) a less than prototypical (see 1.4) status, but this is a minor point, for the properties which the example does possess are highly prototypical for the expression-type in question.
All of exx. (1)-(6) are similar as to their form in that the scalar combination occurs in the typically generic (see 3.1.3) hos te-clause. But the combination 'adjective +per f unctioning as a scalar superlative is by no means confinedto this particular environment. It may occur elsewhere, provided that non-factmodality obtains:
(7) (The hero Diomedes jumps down from his chariot in order to fight.This is a terrible sight:)hupo ken talasiphrona per deos heilen"The stoutest heart might well have been dismayed." (II. 4,421).
(8) (There is a fierce battle around the corpse of Sarpedon, king of the
Lycians:)oud' an eti phradmôn per anêr Sarpêdona dionegnô, epei beleessi kai haimati, kai koniêisinek kephalês eiluto diamperes es podas akrous"The sharpest eye would now have failed to recognize the admirableSarpedon, completely covered as he was from head to foot withweapons blood and dust." (II. 16, 638).
These examples have all the pertinent properties in common with (1)-(6). Theirnon-factual modality is not due to genericness, but to 'potentiality': (7)-(8) are
concerned with what might have happened on some particular occasion (seefurther 3.1.3). Their function in the discourse is similar to that of (1)-(6), too:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 81/318
SCALARITY AND THE PARTICLE PER 71they are used for the statement of a superlative fact. In (7), Diomedes' jumping from his chariot is presented as something in the extreme, and in (8) the
battle around the corpse of Sarpedon is not negligeable either.
3.1.2 Scalar superlatives and the etymology of per
The examples presented above, especially ex. (6), lend themselves easilyfor an analysis in which per is equated with a focus particle in a modern Westernlanguage. Accordingly, they are the paradigm cases in Fraenkel's (1925) account of per, in which per is equated with auch (see 1.3.3). As I said, there ismuch that is good in this approach, its most important merit being that it introduces the semantics of focus phenomena into the study of per. But the directequation of per and auch is clearly not the ultimate description of per in (1)-(8). Among other things, it totally ignores the intensional character of the focusconstituent of per.
One of the immediate consequences of Fraenkel's account, in which heexclusively focuses on per, ignoring the properties of the contexts in which itoccurs, was that it made him reject the traditional etymology of per (see 1.3.4),
in which the original meaning of per is considered to lie in the intensifying('superlative') sphere.
I think that the description of (1)-(8) which I have presented above makesthe conflict between the alleged original meaning and Fraenkel's centralsynchronic meaning less flagrant. This is due to the introduction of inten-sionality into the discussion of scalarity and focus phenomena. The correct account of the scalar superlative involves precisely the intersection ofintensionality and scalarity. It combines the adjectival property of superlative-ness with the semantics of focus and scalar phenomena.
From this double nature it follows that (1)-(8) and the use of per thereincan be approached from two directions: one can focus either on the adjectiveor onper. In the former case the result is an analysis in terms of intensification,and in the latter case it is difficult to escape Fraenkel's equation, or somethingsimilar to it. The two accounts are conflicting, according, to Fraenkel, but theyyare conflicting because they are incomplete. The conflict disappears when theadjectival, intensifying aspect and the scalar aspect are seen as complementary.They are the two inherent properties of the scalar superlative as this phenomenon is discussed in the previous chapter.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 82/318
72 LINGUISTICS AND FORMULAS IN HOMER
3.1.3 Epic te and generic modality
As has been stated (2.1.4), intensional scalarity involving scalar superlatives is distributionally confined to irrealis modal contexts. In ex. (7) and (8),the non-factual modality is due to the particle ken or an, which, in combinationwith a verb with past indicative tense-modality marking, denotes what mighthave occurred in some situation, in other words, a possibility in the past (seeKühner-Gerth i: 212). The non-factuality lies in the fact that Diomedes jumping from his chariot being feared by someone and Sarpedon being recognized
by someone uninvolved in the fighting remains a mere possibility.In (1)-(6), on the other hand, we have to do with facts. But the facts in
volved are independent of actual times and circumstances. The genericmodality resulting from this independence is overtly expressed by a curious linguistic phenomenon, which is, just as the use of per under study, confined toearly epic, viz. the adverbial ('modal') use of the particle te.
Te is etymologically connected with Indo-European k we (cf. Latin -que);as such its function is primarily connective (A Bte, 'A and B'). However, inHomeric Greek it has yet another function. Here it may signal, as a sentential
adverb, what Ruijgh (1971) calls a 'permanent fact' ('fait
permanent'). Clausesin which this 'epic' te occurs are statements about 'permanent' (generic) factsof the world as the poet and his audience know it. Such permanent, generic,truths are not likely to make an essential contribution to the flow of discourseor narrative. Timeless truths are more suitable to function as a 'digression', asa backgrounded statement, which may serve as a 'background' against whichwhat is narrated may be understood or assessed. Consequently, the natural environment for adverbial te is the non-restrictive ('digressive') relative clause(Ruijgh 1971:2). This kind of clause typically contains backgrounded information, which is not essential for the identification of the extension (see 2.2.2) of
the (head) noun to which the relative clause is attached.8
The backgrounded, digressive information expressed in the relative clauseintroduced by the combination 'relative + te' (hos te) may be less essential fromthe point of discourse, but from the point of view of ontology or biology itascribes an essential property to the extension of the head noun. What is saidin the te-clause is often so essential and conspicuous a property of somethingthat it is the most characteristic thing that can be said about it. The te-clauseexpresses the very nature of the referent. An example:
(9) all' agorêtai
esthloi, tettigessin eoikotes, hoi te kath'hulên
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 83/318
SCALARITY AND THE PARTICLE PER 73
dendreôi ephezomenoi opa leirioessan ieisi(The elder Trojans: "but they were excellent speakers, Hke cicadas,which, sitting on a tree in the woods sing delightfully").
Anyone who has been in a Mediterranean wood in summer knows that the relative clause in (9) says something very essential and characteristic about cicadas.The essential character Ues in the fact that the te-clause is not true of some particular group of cicadas, but of all (kinds of) cicadas, of the cicada as species.
What a cicada does qua cicada is sit on aa tree and chirp continuously. Thus thehead noun of the te-clause exemplified by (9) refers to types (species, classes,concepts), rather than to tokens (individuals, particular instances).
Now what is true of a type is of necessity true independently of particularcircumstances. From the point of view of human perception and cognition, theattributes and properties of species, classes and concepts are entirely unchanging and stable. They axe generic properties in the full sense of the term. Accordingly, the propositions in which these generic/essential properties areexpressed are, too, generic in the fullest sense: they involve the complete ab
sence of temporal reference (see Lyons 1977: 194). Propositions in which properties are ascribed to types are true of all times. Thus the te-clause involves the strongest form of generic modahty, viz. timelessness. Notice that lessstrong cases of genericness (see 2.1.4 above) consist in the absence of specifictemporal reference rather than in the absence of temporal reference as such.10
The hos te-clause provides a very suitable environment for scalar superlatives. As stated above (3.1.1), most instances by far of the combination 'ad
jective + per' functioning as a scalar superlative occur in a hos te-clause. Ascalar superlative may occur in the hos te-clause simply when the essential,characteristic property happens to be something in the extreme. Thus in (1)-
(6) timelessness and superlativeness go hand in hand. The 'timelessness' is anessential property of the entity to which the head noun refers. The Earth, the
Netherworld, anger, wine, Phobos (hypostatized panic) are entities belongingto all times. And if they are timeless by their very nature, so are the essential
properties ascribed to them.
3.1.4 A remark on diachrony
The digression on epic te has yet another justification beside the fact thatthe scalar combination 'adjective + per' has a marked affinity with the hos te-
clause. The systematic co-occurrence of the use of per in question with 'epic' te
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 84/318
74 LINGUISTICS AND FORMULAS IN HOMER
allows us to make some inferences about the diachronic status of the scalarcombination 'adjective + per'.
Per and te as they are used in exx. (1)-(6) are peculiarities of early(Homeric) Greek. Admittedly, epic te expressing a 'permanent fact' does occurafter Homer (see Ruijgh 1971: ch. 30-2). But in these cases we have an overt
poetic ( = 'epicizing') language use which plainly differs from contem poraneous ordinary language. As for per, the use exemplified by (1)-(8) simp
ly does not occur after Homer.In section 1.5 above it was stated that the stage in the development of the
Greek language represented by Homer is by no means synchronicallyhomogeneous. On account of the preserving influence of formulas, many linguistic phenomena that have long disappeared from the ordinary languagekeep playing an important role in the epic, oral-formulaic diction. But this
preserving tendency does not preclude a continuous process of renovation, aconstant influx of contemporary linguistic data. From the point of view of
phonology and morphology, this process has been extensively documented (seeJanko 1982).
But the typically epic diachrony applies to semantic phenomena as well.The formulaic preservance of obsolete meanings of words or phrases, togetherwith the admission into the diction of more recent uses may lead to an unnatural, typically epic, co-occurrence of diachronically heterogeneous uses, a
phenomenon which deserves of as much attention as the epic co-existence inthe case of morphological and phonological items.
The particle te is an acknowledged case in point. To all appearances, theuse of te in sentences ascribing some essential (generic) property to a class orconcept is in Homer's time not a synchronically productive idiom anymore. Its
meaning has been subject to attrition, to the effect that a number of instancescan be found in which te has no recognizable meaning. These examples can
be seen as 'the final stage of a process' in which the adverbial ('modal') use ofthe particle gradually faded (see Hoekstra 1981: 14; cf. Ruijgh 1971: 12). Terobbed of its meaning came to be used for merely metrical and/or formulaic
purposes.It seems now a plausible step to apply the diachrony of te to the use of per
in question. We saw that the distribution of the scalar combination 'adjective+ per' is almost entirely confined to the 'digressive' hos te-clause. Now when
the typically timeless hos
te-clause can be shown to belong to a stage in thedevelopment of the language prior to the poet's own, it is likely that this is alsotrue of an expression that is more or less systematically connected with the hos
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 85/318
SCALARITY AND THE PARTICLE PER 75te-clause. This seems to be confirmed by the fact that per as it is used in (1)-(8) does not occur after Homer.
The diachronic development of both particles appears to be parallel tosome extent. That te can indeed be seen as a sort of index of the diachronicstatus of instances of per is shown by the following example, in which an instance of an unequivocally recent use of per (viz. per in 'the same as-contexts',see 2.3.2.2 above) co-occurs with an evidently meaningless (and hence recent)case of te. The example is wholly equivalent to exx. (15)-(16) of 1.2:
(10) toios eon hoios essi, ta te phroneôn ha t' ego per"being as you are and thinking those things that I < think >." (Od.7,312).
The second use of te here is totally irregular.15 Being devoid of meaning, itseems to be inserted merely for metrical purposes. The extent of the deviancefrom the original use can, moreover, be gauged from the fact that in occurringin a restrictive relative clause te is dissociated from its natural locus, the non-restrictive relative clause. Thus the distribution ofte and per seems to conform
to a consistent pattern: old co-occurs with old, and recent with recent.3.2 Extensional terms
In the present section we deal with the group of instances which is exemplified by ex. (3) in 1.2 above. Per is here an organic part of a non-restrictive relative clause; it occupies, just as te in the hos te-clause, the positionimmediately after the relative pronoun: hos per. This use, in which per , insteadof having a proper focus constituent, modifies a whole subordinate clause, isnot particularly what one would associate with a scalar particle. Yet the ex
amples in question are closely connected with scalarity, in that the head nounof the relative clause introduced by hos per is the focus constituent of the scalar particle kai (or oude), the Greek equivalent of even. This is why the examplesin question have to be discussed in connection with 'scalarity', rather than fromthe point of view of the relative clause. As stated in 1.1, the use of per in subordinate clauses will not be discussed for its own sake. In the present study Iwill deal exclusively with per in scalar(-concessive) relative and conditionalclauses.
The expression-type in question conforms to the general pattern 'kai +focus constituent + hos per 9. It provides a link between the use of per discussed
in the previous section and the use of per with participles. This link will be fur-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 86/318
76 LINGUISTICS AND FORMULAS IN HOMER
ther documented in 4.2.2, where I will discuss an expression-type in which 'participle + per' comes in the place of the hos per-clause.
3.2.1 The intension of focus constituents
In section 2.2.1 above I argued that the distinction between the referential and the attributive use of noun phrases should not be maintained in therigid form given to it by philosophers (e.g. Donellan 1966). Noun phrases nor
mally perform two functions at once; they point to the referent to which aspeaker wants to refer, and they say about that referent the relevant thing tosay in the particular context or situation. In other words, most nouns or noun
phrases have both an intensional and an extensional aspect. There are somenoun phrases, however, that have to be placed at the extensional extreme ofthe scale, viz. proper names. This kind of nouns typically possesses a minimumamount of descriptive content. Proper names are indeed the most neutral wayto refer to someone.
In being the most neutral way to refer, proper names are also the most
context-dependent way to do so. A speaker referring to someone with a propername usually supposes his addressee to have so much situational knowledge asto be able to identify the person meant without description ('intension'). Thisneutral, intension-less and context-dependent reference has consequences forthe scalar use of proper names. When a proper name functions as the focusconstituent of even, the scale whose high point it represents is clearly scope(context) dependent (see 2.3.1.4 above). Nothing in the meaning or content ofthe focus constituent and its alternatives explains why they have the positionthey have on the scale in question.
Now there is a difference between the simple use of proper names (Paul
Hked the excursion') and the use of proper names as the focus constituent ofeven ('Even Paul Hked the excursion'). In the first case, simple situationalknowledge of the referent of the name (knowing who Paul is) suffices to makesense of the utterance in which the name occurs. In the second case, on theother hand, something more is involved. Paul can only be used as the focus constituent of even by virtue of some superlative property or attribute of Paul. Andthe addressee, beside knowing who is meant by Paul, has to know this property if he is to make sense of the utterance in which Paul occurs, or, more exactly, if he is to make sense of the scalar implicature (see 2.3.1.1) of even. Thus, in
the case of the scalar use of extensional terms like proper names, some intensional aspect does exist, which need not, however, be overtly expressed. In manycases it is understood in the context of utterance.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 87/318
SCALARITY AND
THE PARTICLE PER 77
A pragmatic account of reference involves both reference proper and attribution, both extension and intension. In many cases just knowing which particular referent is referred to is not enough; the question is often, if not always,qua what it is referred to. This applies to an even greater degree to the referential focus constituents of even. The 'qua what' is here all-important. It involvesthe (relevant) superlative property of the referent, the very reason why thefocus constituent can function as focus constituent.
This principle equally applies to the most neutral kind of reference, the
use of purely extensional terms, viz. proper names. Reference with propernames usually suppresses the 'qua what'. However, when a proper name functions as the focus constituent of even, the 'qua what', the relevant superlative
property of the referent becomes a factor determining the appropriateness, oreven the truth, of the utterance in which the proper name occurs. In 'EvenPaul liked the excursion', Paul is not (neutrally) referred to qua Paul, but qua'the least likely person to appreciate the excursion in question', for whateverreason. This is due, as we have seen (2.3.1.1) to the conventional scalar implica-ture of even.
When the addressee knows, and is supposed to know, that this particularsuperlative attribute can be ascribed to Paul, nothing is amiss. But when theaddressee does not know the relevant 'qua what' of Paul in the particular situation, or may be supposed to miss it, the utterance as it stands may be consideredinfelicitous. If it is to be a wholly meaningful (viz. appropriate) utterance, thereference to Paul must be completed. In other words, the superlative intension,
by virtue of which Paul can be used as the focus constituent of even has to be provided. The hos per-clauses to be discussed in the present section have precisely this function. They complete the reference of a focus constituent ofeven, thereby explaining the scalar implicature of the particle.
3.2.2 Superlativeness, comparativeness and the illocution of even
The raison d'être of the following relative clauses introduced by hos per isthe fact that their head noun is a proper name which functions as the focus constituent of kai or oude ('(not) even'. The relative clause each time providesthe relevant intension of the head noun, the superlative property by virtue ofwhich the head noun functions as focus constituent:
(11) (The Trojan Helenus contends that Diomedes is the most
redoutable of the
Greeks for the Trojans:)oud' Akhilêa poth hôde g' edeidimen, orkhamon andrôn,
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 88/318
78 LINGmSTICS AND FORMULAS IN HOMERhon per phasi theas ex emmenai."We never were so terrified even of Achilles, leader of men, who issaid to be a goddess' son." (II. 6,1000).
(12) (Achilles speaking: death is wholly inescapable.)oude gar oude biê Hêraklêos phuge Kêra,hos per philtatos eske Dii Kroniôni anakti."Even the mighty Heracles did not escape his doom, dearest as he
was [who was dearest] to Zeus the royal son of Cronos." (II.18,118)(13) No one and nothing has the power to resist Atê ('Blinding'):
kai gar dê nu pote Zên' osato, ton per ariston
andrôn êde theôn phas' emmenai."Why, even Zeus was blinded by her once, and he is said [who is said]to stand above all men and gods." (iZ. 19,95).18
(14) (Achilles's horse speaking:)noi de kai ken hama pnoiêi Zephuroio theoimen,hen per elaphrotatên phas' emmenai"We may run even with the West Wind's speed - and there is nothingfaster [which is the fastest thing] known on Earth" (- but we will not
be able to prevent your death). (II. 19, 416).
Each of these examples serves in its appropriate context, from a communicative ('rhetorical') point of view, as an argument for some superlative or, whatis in practice equivalent, inescapable fact. As we have seen (2.3.1.2), this is com
pletely in line with the illocutionary potential of sentences containing even:those sentences are typically uttered not for their own sake, but to state, in amore or less indirect way, a superlative fact. In the context of (11), this factis the extreme fear of the Trojans of Diomedes, in (12) the utter impossibilityto escape Kêr, the god of death, in (13) the irresistible power of Atê, and in (14)the inevitable death of Achilles.
The extensions of the focus constituents of kai and oude are highly suitedto this argument, as the relative clauses explain. The superlative property is in(12)-(14) expressed by genuine, morphosyntactic superlatives (philtatos, oris-tos, elaphrotatên, 'dearest', 'best' and 'swiftest' resp.) and in (11) the attributeof being a goddess' son may be considered a 'pragmatic superlative.'20 The su
perlatives may be seen as in index of the typical function of the hos Per-clauses
in (11)-(14).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 89/318
SCALARITY TO THE PARTICLE PER 79
Sometimes the hos per-clause provides a property of the person to whichthe head noun refers which lies in the comparative, rather than in the superlative, sphere. In these cases, a morphosyntactic comparative will occur in therelative clause. This different content is connected with a different communicative (rhetorical) function of the sentence in which the head noun occurs as focusconstituent. Here we have not so much the assertion of a superlative fact as a
persuasive or dissuasive utterance.An utterance in which an even-phrase occurs may very well be used in an
attempt to influence the behavior of an addressee. In terms of speech acttheory, it may have a directive illocutionary force. Since an utterance of thetype 'Even Paul did it' is a less straightforward way of manipulating the behaviorof an addressee than a simple imperative ('Do it'), it may be called an indirect
directive speech act. We may speak of 'indirect speech act' when a sentencehas an illocutionary force which does not agree with the sentence-type to whichit belongs. In the example 'Even Paul did if we have a declarative sentence witha directive illocutionary force, which makes an indirect speech act.
Now when we supply the relevant intension of the focus constituent ofeven in a directive speech act, this intension tends to be in the comparative,rather than in the superlative sphere, and we get a clear case of a fortiori argumentation, which conforms to the following pattern: "Even X...; and X ismore likely to...than you. So you have to..." The following instances of hos perare straight forward examples of the speech act in question and its concomitantargumentation. The relevant intension of the head noun/focus constituent(again invariably a proper name) provided by the hos per-clause explains howthe a fortiori works:
(15) (Agamemnon to Menelaus: 'Withdraw, don't fight Hector!':)kai d' Akhileus toutôi ge makhêi eni kudianeirêierrig' antibolêsai, ho per seo pollon ameinôn."Even Achilles fears to meet him in the field of honour, and Achillesis [who is] a better man than you by far." (/7. 7,114).
(16) (Apollo to Patroclus:)khazeo, Diogenes Patrokleis: ou nu toi aisasoi hupo douri polin perthai Trôôn agerôkhôn,oud' hup'Akhillêos, hos per seo pollon ameinôn."Back, my lord Patroclus! The city of the lordly Trojans is not
destined to be captured by your spear, nor even by Achilles, who isa better man than you." (II. 16,709).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 90/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 91/318
SCALARITY AND THE PARTICLE PER 81
The predicate of the scope sentence ('be afraid') and the relevant intension of the strongest item substitutable for the variable (Very brave') may beseen as opposites. The opposition is always present; it can be made visible bythe addition of a constituent which explains why the focus constituent is theleast likely value substitutable. Whenever this explicatory constituent is added,it acquires the semantic value of a concessive adverbial element.
Thus we see that the often noticed concessive meaning of per is not aseparate 'sense' of the particle, of which it may be asked whether it has any
thing in common with other senses. 'Concession' is the immediate consequenceof the superlative intension of the focus constituent. And as per is the particleof intensional scalarity, it is clear that the connection between scalarity and concession is at the very core of the meaning of per. In the following chapter (4.3)I will show that the semantic connection between scalarity and concession mayturn into a transition: the meaning of per has been subject to a diachronicdevelopment in which scalarity gradually disappeared and in which concession
became a factor in its own right.
3.2.4 Scalar expressions and the Homeric narratorIn the present (3.2) and the previous (3.1) section we have seen that per
plays an essential role in two scalar expression-types which express a superlative fact. Per either explicitly marks an adjective as a scalar superlative, or itmarks the appropriate and relevant intension of an extensional term which isused for scalar purposes. In itself the distinction between intensional and extensional scalarity is clear enough. But it does not reveal why in a given contextthe intensional scalar expression is preferred to the extensional one. In otherwords, it does not explain the relative distribution of the two scalar expressions.
In the present subsection I suggest an explanation for this distribution in thetext of Homer.It is a matter of simple observable fact that all of the examples (11)-(18)
cited in 3.2.2 occur in passages consisting of interactive speech. In the case ofcomparative intension ((15)-(18)) this observation has already been made ex
plicitly. A directive speech act, which is meant to influence the behavior of anaddressee (or, for that matter, any speech act), must of necessity occur in interactive speech. But the examples in which superlative intension is involved((11)-(14)), too, occur in exchanges between personages. The scalar superlatives cited in 3.1.1 (exx. (1)-(8)), on the other hand, invariably occur in narra
tive passages. This exceptionless assignment to two clearly differentiated(con)text-types seems to be in accordance with the nature of the two expres-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 92/318
82 LINGUISTICS AND FORMULAS IN HOMERsion-types in question, and it provides a clue as to the explanation of their relative distribution in the text of Homer
Someone who uses an utterance in which 'even x' occurs as an argumentfor ( = as an indirect way of asserting) a superlative fact, for example,
(20) Hector is extremely redoubtable: even Achilles fears to meet him,
normally commits himself to some belief or opinion, viz. that the extension ofthe focus constituent, Achilles, is in fact the least likely person to fear Hector.
This commitment is nothing but the conventional scalar impUcature which isinherent in the use of even (see 2.3.1.1 above). In the case of intensionalscalarity, on the other hand, such a commitment is absent:
(21) Hector is extremely redoubtable: the bravest warrior would fear tomeet him.
Here the scalar phrase is non-referential, like any scalar superlative. There issimply no person a belief about whom can be implicated at all. Consequently,the argument in (21) for Hector's supreme skill in combat is, on acount of the
absence of the implicature, much more neutral than in (20). Now neutrality in the sense of not expressing (by whatever means) one's beliefs or convictions about persons is, in general, a characteristic property ofthe narrator in the Homeric poems. From the point of view of narratology, thestory of the Iliad and the Odyssey is told by an omniscient narrator, who doesnot himself play any role in the action he reports. Yet in spite of his omniscience he does not normally impart to his hearer his opinions and beliefs aboutthe elements of his story (persons, events) in any direct way. He purports toreport the chain of events of which his story consists neutrally and objectively.His narrative aims may be said to consist in showing rather than in telling,29 by
speaking as little as possible in his own voice.The obvious method to achieve this aim is to present things happening not
from the point of view of the narrator, but from the point of view of the characters in the story, that is, to introduce large amounts of direct speech.31 In directspeech the narrator is not telling what happened: he presents things as they ap
pear in the perception of one of the characters;3 what is said is put on accountof the one who is speaking, not on that of the narrator.
The nature of Homer's narrative technique is a priori likely to have linguistic implications on discourse level. The language use of the Homeric nar
rator obviously differs considerably from the language use of the characters inthe story. In the two scalar expression-types under study so far in this chapter,
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 93/318
SCALARITY AND THE PARTICLE PER 83
exemplified by (20) and (21), we have now an example of a linguistic phenomenon whose distribution in the text seems to be determined by the narrative technique employed. The scalar implicature of even, which involves thecommitment to a belief or an assumption about the referent of the focus constituent, is ill compatible with the external and detached persona of theHomeric narrator. Consequently, we find it only in speeches. Intensionalscalarity, on the other hand, does not, and cannot, involve a commitment to a
beUef about a referent. This seems to explain its absence from the speech seg
ments of the Homeric text and its exclusive use by the Homeric narrator.
3.3 Scalarity and comparatives
In the present section I discuss comparative expressions in which thesecond argument of the comparative predicate (see 2.3.2.1) is the focus constituent of even. A simple example of such a sentence is:
(22) Paul is even more clever than BILL.
Sentences like (22) are eminently suited for the assertion of superlative facts*
the normal meaning of even and that of the comparative work together here.It is asserted that the person named Paul has some property (cleverness) to ahigher degree than someone who has this property already to a very high degree: in (22) it is implicated by even that is the least likely person to beless clever than someone else. Yet he is less clever than Paul, and it is thisremarkable fact that is stated in (22). The focus constituent of even here is 'Bill',the second argument of the comparative predicate, or, alternatively, the sub
ject of the comparative sentence (see 2.3.2.1). And the scope sentence of evenin (22) is
(23) Paul is more clever than x.
Now this analysis of (22), in which there is a direct relation between evenand Bill, which has to be stated in terms of focus constituent, is not yet the fulldescription of the scalarity of (22). 'Bill' is an extensional term; and like all ex-tensional terms it has to be marked for scalarity by the insertion of even, as wehave seen (2.2.3). But the attribute of 'being more clever than B i l l ' is of coursean intensional affair; it denotes a property, rather than an actual person orthing. And on account of even the predicate denoting this property is a scalarterm no less than the focus constituent 'Bill' is one: 'being even more cleverthan Bill' implies 'being more clever than persons other than Bill'.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 94/318
84 LINGUISTICS AND FORMULAS IN HOMER
Thus the scalar statement (22) yields two separate scales. The one is ex-tensional and scope-dependent (see 2.3.1.4). It is the scale of the probability of
being less clever than someone else. Its high point is represented by 'Bill'. Theother scale is the opposite: it is intensional and scope-independent; its high
point is the property of being more clever than Bill, the one who is least likelyto be less clever than anyone else. And the person to whom this superlative
property applies is Paul, and not Bill. This double scalarity seems to be confirmed by the fact that in sentences like (22) there is not one stressed constituent ('Bill', the focus constituent of even), but two: the comparative itself(i.e. the adjective) tends to attract emphasis too.
The analysis of sentences like (22) presented here, in which two scalesoperate simultaneously, seems to provide the right explanation for the use of
per in the following examples. Kai here is used in the same way as even in (22):it marks an extensional term which is the compared argument of a comparative sentence, like 'Bill' in (22). Per, on the other hand, is attached to the com
parative particle ê ('than'). It is used to express the scalar properties of thecomparative attribute, and to mark the second, intensional, scale:
(24) (Nestor to the quarreling Agamemnon and Achilles:)alla pithesth': amphô de neôterô eston emeio:êdê gar pot' egô kai areiosin ê per huminandrasin hômilêsa, kai ou pote m ' hoi g' atherizon."Listen to me. You are both my juniors. And what is more, I havemixed in the past with even better men than you and never failed tocarry weight with them." (II 1,260).
(25) (Penelope: 'my son, who is still so young and unexperienced, has
sailed away:)tou dê egô kai mallon oduromai ê per ekeinou."And for him I grieve even more than for his father." (Od. 4, 819).
This use of per, which is not very common in Homer (see the Appendix), wasalready correctly characterized in Ebeling (1880: 163),36 albeit not in scalarterms. It should be noted that not all instances of ê per (or ê per) are scalar like(24)-(25). But these exceptions do not detract anything from the description
presented here. They will be discussed in 7.3.1.We close this section with a unique example of per, which may be called
a context-dependent comparative:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 95/318
SCALARITY AND THE PARTICLE PER 85(26) (Priam is in the tent of Achilles, emploring Achilles for the corpse
of Hector:)all' aideio theous, Akhilleu, auton t' eleêson,mnêsamenos sou patros: ego d' eleeinoteros per"Achilles, fear the gods, and be merciful to me, remembering yourown father. But I am even more entitled to compassion." (II.24,504).
Here the second argument of the comparative predicate is omitted, there being
only a subject term. When a comparative has only a subject term and the com parative clause is absent, this is always contextually motivated: the extent towhich someone other than the referent of the subject term has the property inquestion is presupposed (knowledge shared by speaker and addressee) in thecontext of utterance. When Priam is speaking in (26), the topic of conversationis the misery of Achilles' old father; accordingly, the comparison need not beexpHcitly uttered. Ex. (26) shows that when a 'context-dependent' comparativeis scalar (in (27) it functions as an a fortiori argument), per is attached to thecomparative adjective.
Ex. (26) belongs to the examples which are cited by Denniston ( 1954:
482) as the relicts of the original 'intensive' meaning of per. But this specialtreatment is unnecessary, just as in the case of the other instances of Dennis-ton's 'intensive use' (see 1.3.4). The example can be completely accounted forwithin the scalar framework. Notice that Denniston's rendering ('far more to be pitied than he ' - emphasis added) is unhappy anyway because it does no justice to the scalar properties of the phrase.
3.4 Negative polar scalarity
3.4.1 Declarative sentences
The instances of per discussed in the previous sections (3.1, 3.2 and 3.3)occur in what may be called declarative sentences. The term 'declarative' ap
plies to a sentence-type and may be opposed to interrogative and imperative.37
Speaking in terms of modality (see 2.1.4), declarative sentences may be eitherfactual or non-factual. Factual declaratives are normal indicative statements.An example of a non-factual declarative in which even occurs is
(27) Even Paul would have liked the excursion.
The difference between this sentence and its factual counterpart 'Even Paul
liked the excursion' is that in the latter it is implied that Paul actually participated in the excursion, while in (27) it is not.39 But this (modal) difference
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 96/318
86 LINGUISTICS AND FORMULAS IN HOMERis immaterial for the declarative status of both sentences as well as for their il-locutionary force (i.e. the use of even). In both sentences the same superlativefact is stated: the excursion in question was very interesting. Thus the distinction between 'factual' and 'non-factual' does not entail the distinction betweendeclaratives and non-declaratives. There are non-factual declaratives.
The kind of scalarity that may occur in declarative sentences was discussedin 2.4.1 as positive polar scalarity. The high point of the kind of scale involvedhere is a positive extreme ('maximum'). We saw that negation in connection
with positive scalarity is immaterial; it just may or may not occur. And in Greek,whether kai or oude is used does not make any difference for the kind ofscalarity involved. When negation occurs, the scalar expression stands outsideits scope, and the scope sentence is equivalent to an affirmative. This is different in the contexts we are about to discuss in the next subsection. Here negation is a commanding factor which alters the nature of the scale involved.
3.4.2 Non-assertive contexts and negative polarity
In declarative sentences something is asserted. Assertions may be posi
tive or negative. In negative assertions information is removed, rather than positively introduced. Here negation may be seen as a non-fact modality. Infact negation as a sentential property is so important that negative assertionsform a distinct group within the class of declarative sentences. The status ofthese negative assertions is mixed: on the one hand they are complete and declarative sentences (used for negative assertions), but on the other hand theyshare an important property with some contexts which are syntactically andsemantically the opposite of declaratives. This property of negative sentencesconcerns their readily admitting negative polarity items.
Negative polarity items (NPI's, see 2.4.2 above) occur, roughly speaking,either in main clauses in which negation is a commanding feature or in a num
ber of dependent subclauses. These subclauses include conditionals andrestrictive relative clauses. The common denominator of this kind of subclausesis that they do not have, as declarative sentences do, the properties of completeand independent propositions. In other words, they do not on their own referto a state of affairs and they do not belong to one of the sentence-types declarative, interrogative or imperative.
The subclauses in question may be called 'non-assertive'. Their non-assertive character has to be viewed together with their dependent nature. They as
sert something only in combination with the main clause to which they belong.The non-assertiveness of these subclauses can, within the framework of the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 97/318
SCALARITY AND THE PARTICLE PER
87
present study, be related to the notions of wide scope and scale reversal discussed above (2.3.1.5 and 2.4.2). The wide scope of a scalar expression is simply the consequence of the fact that the subclause in which the scalar expressionoccurs cannot on its own function as a sentence, and consequently not as a scopesentence either.
3.4.3 Scalar superlatives and negative polarity
In positive polarity contexts (see 2.4.1) per may modify an adjective so asto make it a scalar superlative. Scalar expressions occurring in assertive, positive polarity contexts belonging to the declarative sentence-type represent thehigh point (maximum) of a positive scale. In negative polarity contexts, on theother hand, the situation is reversed. Here the high point (or better: low point)denotes a negative, minimum quantity or quality.
Now, it is an important property of positive scales that the lexical value ofscalar superlatives in positive polarity contexts is unimportant. The intensionalscalarity of scalar superlatives in positive contexts works independently of thenature of the property denoted by the adjective. What matters is the positive
superlativeness, which can be expressed by morphosyntactic means. Whetherwe are dealing with negative or positive properties ('clever' vs. 'stupid', 'big' vs.'small') is unimportant. In the case of 'smallest', for example, we are simply ex
pressing the maximum amount of a negative property, smallness. The scale inquestion is simply downward implicative: what is 'smallest' is by definitionsmaller than what is less small.
In the case of intensional scalarity in negative polarity contexts, on theother hand, there is a lexical constraint on scalar superlatives, in that only ad
jectives denoting negative properties are allowed. Adjectives denoting positive properties cannot yield scalar readings in negative environments. The sentence'John does not understand the most difficult author', for example, does notimply that 'John' does not understand less difficult authors. In fact he may verywell understand these. On the other hand, 'John does not understand the least
difficult author' implies John's not understanding any other (i.e. more difficult)author.
The lexical constraint on scalar superlatives in negative contexts can beunderstood as the combined operation of scalarity and negation. When eventhe smallest quantity/amount is not the case, then it is clear a fortiori that anylarger quantity/amount will not be the case either. Thus in the case of scalarity
in negative contexts we are dealing with minimum amounts of negative properties. Consequently, we may say that the scale associated with this kind of scalar
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 98/318
88 LINGUISTICS AND FORMULAS IN HOMER
expressions is reversed with respect to the corresponding expressions in positive contexts.
We now apply this negative polarity/scalarity apparatus to Homeric per .The basic meaning of the particle emerged so far can be stated in terms of intensional scalarity. On the basis of this fact we are in a position to predict thatwhen per occurs in a negative polar context, its intensional focus constituentmust be an adjective denoting something negative. This prediction is borne out
by the facts. These facts are not numerous -per occurs no more than three timesin a negative polar context, with wide scope - but they are sufficient. Each time
per modifies oligos ('small', 'little'):
(28) (Diomedes to Paris: 'You are a weakling and your shot did me noharm'.)ê t' alios hup' emeio, kai ei k' oligon per epaurêi,oksu belos peletai, kai akêrion aipsa tithêsi."My weapons have a better edge. One touch of them and a man isdead [Ut: even if they touch a man ever so little]." ( . 11, 391).
44
(29) anti kasignêtou kseinos th ' hïketês te tetuktaianeri, hos t' oligon per epipsauêi prapidessin."To any man with the slightest claim to common sense, a strangerand a suppliant is as good as a brother." (Od, 8,547).
(30) (Aeolus, the king of the winds, lends Odysseus a helping hand. Hehas emprisoned the energy of all the winds in a big leather bag:)nêi d' eni glaphurêi katedei mermithi phaeinêiargureêi, hina mê ti parapneusei' oligon per.
"This pouch he stowed in the hold of my ship, securing it with a silverwire so as to prevent the slightest leakage." (Od. 10,24).
These examples present three different negative polarity contexts. In (28)we have a conditional and in (29) the kind of restrictive relative clause whichis sometimes called 'conditional', because it allows of the paraphrase by a conditional ('if someone'). Both clauses are dependent on their main clauses;they do not on their own express a complete proposition. Clausal dependenceis expHcitly marked in Greek by the subjunctive, the subclause-mood par excellence. And it is the subjunctive that we see appear in (28) and (29). In (30),
the negative polarity is due to negation in a final subclause. Negated final subclauses are as regards the acceptance of negative polarity items equivalent to
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 99/318
SCALARITY AND THE PARTICLE PER 89the complement of negation-implying verbs like fail, refuse and prevent. See thetranslation of (30).
Just as in the case of scalar superlatives in positive, assertive, contexts,EngUsh is instructive in that it has a means other than an expression with evenor other scalar particles to express the scalarity, viz. the morphosyntax of thesuperlative. The slightest is morphosyntactically the superlative of slight. Seman-tically and distributionally, however, it is a negative polarity item, expressingscalar superlativeness in contexts where a constraint on the lexical value of the
superlative obtains. The same appHes to oligon per. The adjective is marked by per for scalarity, just as any adjective may be marked so. The rest of the job isdone by the lexical value of oligos, on account of which oligon per can occur asa scalar superlative in a negative polar context and as such be equivalent to anegative polarity item.
3.4.4 The strengthening of negation
Negative polarity items may be discussed in terms of negation strengthening (see Horn 1978a: 148 ff.). In (30), for example, the expression containing
oligon per ('so as to prevent the slightest leakage' is stronger than 'so as to prevent leakage'. One special group of negation strengtheners consists of ex pressions like (not) at all, (pas) du tout (French), helemaal (niet) (Dutch). Aspecial place in this group occupies German gar (nicht). This strengthenednegation may be analyzed as a scalar particle which takes the negative elementas its focus constituent. The use of the scalar particle gar as a negationstrengthener seems to be the appropriate background for the assessment of thefollowing examples, in which the addition of per to the negative particle has astrengthening effect on the latter:
(31) (Aias has struck Hector with a stone:)hôs d' hoth ' hupo plègês potros Dios ekseripêi drus
prorrizos, deinê de theeiou gignetai odmêeks autês, ton d' ou per ekhei thrasos hos ken idêtaieggus eôn, khalepos de Dios megaloio keraunos."(Thus Hector in his gallantry was brought down in the dust) by assudden a stroke as that of Father Zeus when he uproots an oak; itgives off an appalling reek of sulphur and he who is seeing it fromnearby is scared [has no confidence at all], for terrible is the lightning
of Zeus." (II. 14,416).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 100/318
90 LINGUISTICS AND FORMULAS IN HOMER
(32) (Odysseus to the Phaeacians in the Games: Ί will not challenge myhost Laodamas'ton d' allôn ou per tin anainomai oud' atherizô,all' ethelô idmen kai peirêthêmenai antên."But of the rest of you there is really no one [no one at all] I'm too
proud to take on; in fact I am ready to meet and match myself againstall comers." (Od. 8,212).
These examples belong, just like ex. (26) above, to what Denniston ( 21954:482)has called the intensive use of per, which allegedly stands apart from the otheruses. But (31) and (32) can be perfectly dealt with in the negative polarity areaof the scalar framework. A special treatment is unnecessary. Notice that perused as in (31)-(32) does not occur after Homer. Later Greek uses compoundsof the type oud-(h)amos ('not in any way') for the strengthening of negation.
3.5 Scalarity in wishes and commands
3.5.1 Imperative and optative sentences
We have now dealt with per in positive polarity (assertive) and negative polarity (non-assertive) contexts. The former belong to the declarative sentence-type, whereas the latter do not belong to any sentence-type at all, because they are syntactically dependent. We now go on to discuss the use of perin another sentence-type, viz. the imperative sentence. This sentence-type hasa predicate in the imperative mood; it is, accordingly, very suited for the utterance of commands. It will appear in the present section that as regards theuse of per there is no difference between commands and wishes. The expression-type in question, then, involves both the imperative and what may be
called the optative sentence-type.Optative and imperative sentences are non-factually modalized. Their
non-factuality is closely connected with the modality inherent in the futuretense paradigm: when we utter a wish or a command, we are saying somethingwhich is related to a state of affairs that does not (yet) obtain. It is the samekind of non-fact modality as in the complement of verbs like want, strive etc.(see note 20 of ch. 2).
The difference between non-fact modality in negative polar contexts andnon-fact modality in optative and imperative sentences Hes in the fact that the
latter may be conceived of as the modification of a complete proposition. Acomplete proposition (or predication) may be defined 49 as a predicate with allits argument-slots filled by the appropriate terms. The three-place predicate
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 101/318
SCALARITY AND THE PARTICLE PER 91give, for example, needs three noun phrases to form a complete proposition/predication (Agamemnon gives Achilles a woman). This predication maythen be modalized according to the tense or mood marking of the predicate:factual modality obtains in the case of present or past indicative tense-moodmarking, and non-fact modality in the case of future indicative, optative or im
perative tense-mood marking. Non-fact modality may be expressed by mor- phosyntactic means or by auxiliaries or both.
In negative polar contexts, on the other hand, the non-fact modality is not
due to the tense-(aspect-)modality marking of the predicate, but to negation(whether overt or covert). The difference between optative/imperative irrealiscontexts and negative polar contexts appears from the fact that negative
polarity items are distributionally excluded from optative and imperative sentences, as a simple test will show.
Negative polarity contexts have their own kind of scale, which is a reversedversion of the 'normal' scales associated with assertive contexts and the declarative sentence-type, as we have seen. Optative and imperative sentences, too,have their own kind of scale. This kind of scale, which is closely associated with
the specific non-fact modality of optative and imperative sentences, I will callthe desirability scale. Desirability scales mark a number of items by degree ofdesirability in a given context. Now, it is a simple fact that what one desires ina given situation is often more than what is actually possible. As will appear,this common-place observation is the clue to the description and explanationof per as a scalar particle in the optative and imperative sentence-types.
3.5.2 Inclusiveness and exclusiveness
The instances of per under study here form a well-defined group. They are
exemplified by exx. (4) and (5) of 1.2 above. Consider:(33) (Achilles to his mother Thetis:)
mêter, epei m ' etekes ge minunthadion per eonta,timen per moi ophellen Olumpios eggualiksai,
Zeus hupsibremetes, nun d' oude me tutthon eteisen."Mother, since you, a goddess, gave me life, if only for a little while,surely Olympian Zeus owes me at least honour. But as it is, he givesme none." ( . 1, 353).
The first instance f per in this example (minunthadion per ) we pass in silence,so as not to complicate matters more than is necessary. It will be discussed in4.3.3 below. What matters here is timen per. The next example, the end of a
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 102/318
92 LINGUISTICS AND FORMULAS IN HOMER
prayer of Agamemnon to Zeus, contains no less than two instances of per ofthe kind in question:
(34) alLa Zeu, tode per moi epikrêênon eeldôr:
autous dê per eason hupekphugeein kai aluksai,mêd'houtô Trôessin ea damnasthai Akhaious."Ah, Zeus, grant me this prayer at least. Let us escape with our lives,
if nothing else, and do not let the Trojans overwhelm us like this."( 8,242-3).
Other examples:
(35) (Odysseus, member of the Embassy to Achilles speaking:)ei de toi Atreidês men apêkhtheto kêrothi matton,autos kai tou dora, su d' allous per Panakhaiousteiromenous eleaire kata straton, hoi se theon hosteisousi."But if your hatred of Atreides, gifts and all, outweighs every otherconsideration, do have at least some pity on the rest of the unitedAchaeans, lying dead-beat in their camp. They will honour you likea god." (IL 9,301).
(36) Nestor to Patroclus: 'Maybe Achilles has some compelling reasonnot to fight':)alla se per proetô, hama d' allos laos hepesthô
Murmidonôn, ai ken ti phoôs Danaoisi genêai."But let him at least allow you to take the field with the
Myrmidonian force at your back, and so perhaps bring salvation tothe Danaans." ( . 11,796 (cf. ex. (5) in 1.2 above).
(37) (Hypnos, the god of Sleep to Poseidon:) prophrôn nun Danaoisi, Poseidaon, epamune,kai sphin kudos opaze minuntha per, ophr' eti heudei Zeus."Poseidon, you may help the Danaans with all your heart and givethem the upper hand, if only for a short time, as long as Zeus issleeping." ( 14,358).52
(38) (Menelaus rebukes Eteoneus for leaving strangers waiting outside:)ê men dê noi kseinêïa polla phagonteallôn anthrôpôn deur' hikometh', ai ke pothi Zeus
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 103/318
SCALARITY AND THE PARTICLE PER 93
eksopisô per pausêi oïzuos."Think of all the hospitality that you and I enjoyed from strangers
before we reached our homes in the hope that Zeus would spare usat least in the future such pressing need." {Od. 4,35).
In each of these examples, the speaker utters what may be called a modestwish. What he wishes is a kind of compromise, the most desirable thing to bewished for in the context of utterance being left out of account. At first sight,
this property implies a considerable difference between (33)-(38) and the examples discussed earlier in this chapter. The meaning of per in sections 3.1through 3.4 above has very much to do with superlativeness, and it has to bedealt with in connection with even. In exx. (33)-(38), on the other hand, perseems to be different: in accordance with the modest nature of the wish in question, it seems to exclude those alternatives for the focus constituent whichdenote a higher value on the scale. In (38), for instance, the 'present' seems to
be excluded. Likewise, in the other examples exclusion seems to obtain:'honor (and not a long life)' in (33); 'the other Greeks (and not Agamemnon)'
in (35); 'you (and not Achilles)' in (36) etc.The distinction between inclusive and exclusive focus particles (see 2.3.1.3above) may be used for the delimitation of scalar phenomena in language. Allscalar particles are inclusive by their very nature: their focus constituent represents the highest value, and as such it implies (includes) all other (lower)values. Exclusive particles, on the other hand, are not scalar. 'Exclusion' as a property of a focus particle precludes scalarity, because the focus constituentof the particle either does not represent the high point on the scale, if there isone, or it does represent the highest value, but it excludes the lower (i.e. morelikely) ones.55 In either case scalar relations between the focus constituent and
its alternatives are impossible.Thus to say that per in (33)-(38) is exclusive has important consequences,
since it implies a serious disconnection in the array of uses of per. Per wouldturn out to be at the same time a scalar and a non-scalar particle, and the sense-relation between (33)-(38) and the instances discussed earlier in the presentchapter would turn out to be one of homonymy, rather than one of polysemy,there being no relevant common denominator (see note 15 of ch. 1). But a disconnection of this kind can only be acceptable when we have fully applied the
principle of unification (see 1.3.2), viz. when we have made sure that there is in
fact no common property. In the present case this means that we have to lookfor a scalar reading of per in (33)-(38) as long as possible. The present sectionis concerned with this search.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 104/318
94 LINGUISTICS AND FORMULAS IN HOMER
3.5.3 Scales of desirability and acceptability
Per in (33)-(38) can be appropriately rendered by at least in English, asappears from the translations of the examples. For a correct understanding of
per in (33)-(38), we have to realize that the meaning of at least differs according to the modal context in which it is used: in factual contexts ('John has atleast three dogs') it is not scalar in the sense used in the present study. In non-factual contexts (viz. wishes and commands), on the other hand, it certainly isscalar. The description of the scalarity of at least in wishes and commands runs
parallel with that of per in (33)-(38), for here, too, the wish/command character is a commanding feature.
You can't always get what you want. And it is often useless to wish whatis beyond the reasonably possible. This platitude may be reformulated as follows in more appropriate terminology. The maximum fulfilment of one'swishes in a particular situation may be impossible, blocked, as it were, by realityand factuality. In such a situation the non-factual space within which wishes andcommands make sense is narrowed down, and one has to resort to the second
best alternative, as a kind of compromise, so to speak. This second best alter
native may be encoded linguistically as the focus constituent of a focus particle. Now the phrase 'the second best alternative' is clearly suggestive of a scale.But how are we to account of the scalarity and how are we to conceive of thescale involved? As a first try, we might construe a desirability scale, in which theitems are ordered by degree of decreasing desirability in a given context, thehigh point of this scale being the most desirable item. But on this scale thesecond best alternative represents the highest item but one, and it has, as such,exclusive value. Presented schematically:
The phrase which has as its focus constituent a term which represents the
value "B" on this scale cannot possibly be called 'scalar' in the sense of the present study. For one thing, "B" is not the high point, and for another thing,
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 105/318
SCALARITY AND THE PARTICLE PER
95it has no implicative force: if someone wishes (Abeing impossible and, hence,excluded), it does by no means follow that he abo in the sense f α fortiori wishesthe less desirable through F. Consequently, if (39) were the appropriate scalefor the use of per in (33)-(38), we would have to conclude that per is not scalarhere.
However, per and at least in (33)-(38) do have a scalar meaning. But thedesirability scale (39) is not the appropriate way to bring this out. To describeat least and per in wishes as a scalar particle, we have to conceive of a scale of
which the second best alternative "B" of scale (39) is the high point. Such ascale can indeed be construed, and it is closer to the pragmatics of wishes thanappears on first sight. In this scale the items are ordered not by degree ofdesirability, as in (39), but by degree οf acceptability or satisfactoriness. Here thehigh point does not represent the most desirable item in a given context of utterance, but the least desirable item which is still acceptable for the speaker,which still satisfies his wish, albeit in a suboptimal way. This scale is a modification of the higher area of the desirability scale (39):
(40) least desirable but still acceptable
A most desirable
This scale is implicative, and its high point "B" is scalar and, hence, inclusive. For if one contents oneself with "B" as the fulfilment of one's wish, onewill a fortiori do so with "A", the most desirable item. This acceptability scaleis wholly comparable to the scales of positive polarity contexts discussed earlier, since superlativeness plays, again, an important role in it. Its connectionwith the better known scalar phenomena discussed earlier in this chapter aswell as in chapter 2 can be brought out in English when the specific character
of the wish or command is expressed not in an optative or imperative, but in adeclarative sentence; even then appears:
(41) Even B, a suboptimal fulfilment of my wish, will do.
Even in EngUsh cannot, as far as I can see, occur in the overt expressionof the wish or command. At least has to be used instead. This combination,which functions as a scalar focus particle, overtly (morphosyntactically) expresses the superlative position of its focus constituent on the scale: least desirableyet still acceptable.
Homeric Greek, on the other hand, appears to use per for the encodingof the satisfactoriness scale (40), as is shown by (33)-(38). In each of these examples, the focus constituent of per denotes the least desirable realization of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 106/318
96 LINGUISTICS AND FORMULAS IN HOMERthe speaker's wish/command that is still acceptable. In (33) the best thing onecan wish, and especially Achilles, is a long life plus honor and fame. But Achilles has had to choose, and he has chosen honor (timê). Since he has been deniedlongevity, mere honor is the best thing he can hope for in this life. But at thesame time honor is the least thing that still satisfies his desires. In (34) the bestthing a Greek on campaign against Troy could wish is return home safely witha lot of booty after the intended sack of Troy. But as it is in the context of utterance of (34), in which the Trojans are clearly victorious and the sack of Troy
is farther off than ever before, it is more realistic for a Greek to tune down hiswishes, and, accordingly, his prayers to the gods, and content himself with themere saving of his life.
In (33) and (34) the relation between the items wished for may be calledadditive, the maximum of desirability being the sum of two items: 'A + B'. Since'A + B' is unrealizable ('A' being impossible), 'B alone' is the second best itemand the least desirable item that is still acceptable. The relation between thedesired items may also be substitutive: the items are not related to each other
by way of number ('A + + ' being more desirable than 'A + B'), but byway of strength ('A' being more desirable than B'). Of course, the scales connected with superlativeness discussed earlier in this chapter, as well as in chapter 2, are all 'substitutive'.
Examples of substitutive scalarity in connection with wishes and commands are (36) and (37). The focus constituent of per here denotes a 'leastdesirable but still acceptable' item which is related to the most desirable itemnot by way of number, but by way of strength (quality or quantity). But there isalso a difference between the two. In (36) the substitutive desirability scale iscontext-dependent in the way mentioned above (2.3.1.4): the focus constituentof per is a personal pronoun, and as such an extensional term. Nothing in its
meaning explains why 'Patroclus' is the least desirable item that still satisfiesthe speaker's wish. In (37), on the other hand, the desirability scale is scope-independent, and, accordingly, the focus constituent o f per is intensional: if onlya short time will do, it is clear that any longer period will do a fortiori.
Thus we see that the distribution of per in wishes and commands is insensitive to the distinction between intension and extension, as per in wishes andcommands may have both an extensional and an intensional term as its focusconstituent. We saw that the distinction between intension and extension wasindispensable for the correct delimitation of per with regard to kai in declarative sentences and positive polarity contexts, kai being the particle of extensional, and per the particle of intensional scalarity. But just as even, kai cannot
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 107/318
SCALARITY AND THE PARTICLE PER 97
be used for the scalarity in wishes and commands. Consequently, it does not play any role in the delimitation of per in these environments.
3.5.4 Per, ge and exclusion
There remains, however, a particle with regard to which per in wishes andcommands has to be delimited. This is ge. The specific difference on the basisof which the delimitation may be made Hes in the distinction between fact andnon-fact, between statement and wish or command. We saw that in (33)-(38)
the one who utters the wish or command has to resort to a compromise becausethe realization of the highest valued item is impossible. Now this impossibilityis a matter of fact and it is the expression of this fact that may involve exclusive-ness ('limitation').
It is in the expression of the fact which impedes the wish of the speakerthat the particle ge may be used. Ge functions as an exclusive focus particlewhich does not impose an ordering on its focus constituent and the alternativesfor it. When it is used in connection with scalar wishes, it takes as its focus constituent one of the components of the highest item on an additive desirability
scale. The resulting expression-type conforms to the following pattern: '"A +B' is the most desirable item. As a matter of fact 'A (ge)' is impossible. Then Iwish that at least 'B (per)' will happen." Two examples of this expression-type:
(42) (Menelaus to Aias, fighting over the body of Patroclus:)speusomen ai ke nekun per Akhillêï propherômengumnon: atar ta ge teukhe' ekhei koruthaiolos Hektor."We might at least save the body for Achilles, naked though it is.The armour at any rate is in the hands of Hector with the glancingHelm." (II. 17,121).
(43) khaire pater kseine: genoito toi es per opissôolbos: atar men nun ge kakois ekheai poleessi."Your health, my ancient friend! You are under the weather now,
but here's to your future happiness! [May you be happy at least in
the future. For at the moment you are in big trouble]." (Od. 18,122(= Od. 20,199)).
Per and ge in these examples are clearly and unequivocally differentiated: peris inclusive, ge is exclusive, and per is non-fact; ge is fact. The accounts of exx.
(33)-(38) and (42)-(43) in the descriptive literature on per are unsatisfactoryand incomplete precisely in these two respects: they do not discuss the notions
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 108/318
98 LINGUISTICS AND FORMULAS IN HOMER
of 'inclusion' and 'exclusion' in connection with scales and scalarity, and theyunderestimate (or even ignore) the linguistic relevance of the cognitive notions'fact' and 'non-fact'.57
3.5.5 The diachronic aspect of per in wishes
The description presented here shows that per in wishes and commandsdoes not impair the coherence of per as a linguistic sign and the unity of itsdescription. But per in wishes and commands is nonetheless an expression-type
which stands apart from the other uses. From a synchronic point of view thisappears from the fact that its focus constituent in wishes may be an extensionalterm, whereas in all other uses discussed in the present chapter per may becharacterized as the particle of intensional scalarity. Thus when we considerthe instances of per in the synchronic dimension in the light of PrototypeTheory, the instances of per in wishes and commands are less than prototypical in that they lack the prototypical property of intensionality.
From a diachronic point of view, too, the use of per in wishes and commands stands apart from the other uses. The uses of per in scalar superlatives
(see 3.1) and in scalar relative clauses (see 3.2) stand in what may be called thediachronic main stream of the particle, the gradual conversion of intensionalscalarity into concession. In the following chapter I will describe this processin detail (see 4.3.1). Per in wishes and commands, on the other hand, seems to
be, speaking in terms of main streams, a by-path. It is not diachronically relatedto any other use, and it has left no traces in post-Homeric Greek.
The diachronic status of per in wishes and commands seems to be reflected by the quantitative aspect of this expression-type. Out of the 24 instances oiperin wishes and commands (see the statistics in the Appendix) which we find inHomer, only 4 occur in the Odyssey. This remarkable distribution may of course be due to a difference in style between two poets, but it is tempting to assumea diachronic development here, in the course of which per in wishes and commands gradually disappeared. The much lower frequency in the Odyssey is atany rate in line with the lower frequency in general of old items in this poem.59
But the frequency argument does not stand alone. In 7.2 I will present caseswhich point to the possibility that the typical scalar use of per in wishes andcommands was not fully understood anymore in the poet's time.
3.6 Some unclassifiable instances
The examples discussed in the previous sections belong expression-typesthat are as to their form more or less clearly defined. But natural language facts
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 109/318
SCALARITY AND THE PARTICLE PER 99
often do not let themselves be classified so easily. This means, in the presentconnection, that per may have a scalar meaning outside the expression-typesdiscussed in 3.1 through 3.5. In the present section I present those 'unclassifiable' instances. One might call them 'peripheral' in the sense that they do notconform to the classification into various expression-types (see further 7.1.1).But as to their meaning they are not peripheral at all. Per in (44)-(47) below
per functions on its own as a scalar particle. In the examples to be discussed inch. 7, on the other hand, per is either not scalar at all, or, if it is scalar, is redun
dant with regard to the scalar particle oude (see 7.3.2 and 7.3.3).
(44) (Aias to Menelaus in the battle over Patroclus' body:)ô pepon, ô Menelae diotrephes, ouketi nô elpomai auto per nostêsemen ek polemoio."My friend, Menelaos nursling of Zeus; I am beginning to think thatyou and I will never get home safely from this fight (let alone thecorpse).[I do not expect anymore that we can get even ourselves outof here]"(II. 17,239).
This example is similar to the examples discussed in 3.4.3 in that per and itsfocus constituent occur in a negative environment. Auto per represents the high
point of a reversed scale; a minimal quantity is involved. The difference lies inthe fact that unlike oligon per, auto per cannot be described as an inverse scalarsuperlative.Autô is not a purely intensional term; accordingly, the scale in question is scope-dependent, rather than scope-independent.
(45) (Zeus to Here and Athene, quarelling over the battle:)sphôïn de prin per tromos ellabe phaidima guia,
prin polemon te idein, polemoio te mermera erga.
"But you two were trembling before you even saw the battlefield andits horrors." (II. 8, 452).
(46) (Antilochus is on the run for Hector:)α ll ho g' ar' etrese thêri kakon rheksanti eoikôs,hos te kteinas ê boukolon amphi boessi
pheugei prin per homilon aollisthêmenai andrôn"He turned tail like a wild beast that has committed the enormity ofkilling a dog or the man in charge of the cattle, and takes to his heels(even) before a crowd collects to chase him." (R 15,588).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 110/318
100 LINGUISTICS AND FORMULAS IN HOMER
In these examples per marks a temporal adverbial element for scalarity: evenbefore. Notice that in (45) the scalarity is more pointed and functional than in(46): the former is more in line with the characteristic illocution of sentencescontaining a scalar expression.
(47) (Menelaus to Telemachus about his former comrades in the Trojanwar:)hôn ophelon tritatên per ekhôn en domasi moirannaiein, hoi d' andres sooi emmenai, hoi tot' olontoTroiêi en eureiêi, hekas Argeos hippobotoio."How happy I could be, here in my house, with even a third of myformer estate, if those friends of mine were still alive who died longago on the broad plains of Troy, so far from Argos where the horsesgraze!" (Od. 4,97).
At first sight we have here another instance of the use of per in wishes and commands. But in fact the example is different. Menelaus does not utter a com
promise, thereby expressing as the focus constituent of per the least desirable
item that still satisfies his wish. Instead he says: Ί love my old comrades so muchthat I would have renounced the greater part of my fortune for their Uves'. Per
specifies this 'greater part', marking it for scalarity: 'even two thirds'. Viewed
thus, tritatên per can be interpreted as a normal case of scalarity in a declarative sentence.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 111/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 112/318
102 LINGUISTICS AND FORMULAS IN HOMER
the other hand, may be used to ascribe some context-independent property to individuals (e.g.
traits of character or patterns of behavior).
11) But attributive superlatives are by no means excluded from the hos fe-clause, for ex
ample: II 17, 674-5: aietos, hon rha te phasin oxutaton derkesthai hupouraniôn peteênôn ('the
eagle, which has, as they say, the sharpest eye of all birds').
12) For example, as Ruijgh (1971:1001) notes, of the 18 instances of hos te in Aeschylus
(tragedy, 5th century) only 3 occur in normal spoken text. The other examples (15) occur in lyri
cal (choral) and anapaestic passages where the deviance from the ordinary language of the poet's
time is greatest and where epicisms tend to cluster.
13) The examples are cited in Ruijgh (1971: 405 ff.).14) It might be objected here that the frequent occurrence of 'adjective + pef in the hos
te-clause has no diachronic impHcations, on the grounds that the hos te-clause is one of the few
opportunities for the poet to use scalar superlatives at all. On this account, scalar superlatives
occur so often in hos te-clauses simply because they cannot easily occur elsewhere. This may be
true, but to use 'adjective + per' in a hos te-clause presupposes knowledge of the non-factual
(timeless) nature of that clause, and there we are back again to diachrony.
15) Cf. Ruijgh (1971: 421,423). The first instance of te in the example, on the other hand,
can be interpreted as connexive.
16) We might say, with König (1981:117), that even creates an intensional context, i.e. an
environment in which the substitution of co-extensive terms has truth-conditional consequen
ces. See above, 2.2.2 with note 38.
17) For even and negation see 2.4.1. Notice that even in English is a so-called optional
'neg(ation)-raiser', just as predicates like 'think', 'believe': 'Even Paul did not like the excursion'
is equivalent to 'Not even Paul liked the excursion'. In Greek, on the other hand, even (kai) is an
obligatory neg-raiser: the expression of even in a negative scope-sentence has acquired a lexical
status which is different from even in an affirmative scope sentence: oude vs. kai. On even as a
neg-raiser see Seuren (1985:312-13). On neg-raising in general see Horn (1978b), Seuren (1985:
166-72).
18) Notice that kai in this example, as well as oude in (12), is separated from its focus con
stituent (see also ex. (18) below). In later Greek kai gar may function as a fixed collocation which
means 'for it is also true that..', whereby kai takes a whole sentence as its focus constituent. This
use does not yet occur in Homer.
19) See also Ducrot (1980:16).
20) See 2.1.3 on the superlative nature of proper names like 'Kasparov' and 'Getty'.
21) See Lyons (1977: 745-6), Bolkestein (1980: 31), Leech (1983:106).
22) See Searle (1975), Lyons (1977: 785), Bolkestein (1980: 30-1).
23) The distinction between 'direct' and 'indirect' illocutionary force is but one of the
various (sometimes overlapping) distinctions used for classification in this field. See for an over
view Bolkestein (1980: 28-35).
24) But a superlative intension is by no means excluded in directive even-sentences.
25) Notice that this example is the only one of the group in which the intension of the focusconstituent does not involve a time-stable property: the relative clause says what happened to
Niobe at some particular moment. In (11)-(17), on the other hand, the superlative or compara-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 113/318
NOTES TO CHAPTER 3 103
tive intension is unspecified for time. In practice, as most superlative or comparative intensions
denote time-stable properties, but (18) shows that this is not necessarily the case. What really
matters is, of course, the superlativeness or comparativeness.
26) Two alternate morphs for the expression of the relative subordinator in Greek - hos,
the relative pronoun, and ho, an anaphoric demonstrative pronoun, functioning in Attic as the
definite article - provide the flexibility which is necessary for a formula which has the main
caesura as one of its boundaries. The relative clause introduced by hos per (metrically, before a
consonant:- - ) occurs when the caesura is realized as (P)enthemimeres and may be called, in
the notation of Hoekstra (1965), a P2-formula. The relative clause introduced by ho per (metri
cally: U - ) , on the other hand, occurs when the main caesura is realized as T (trochaic caesura).
It is a T2-formula. See further on these matters (i.e. the interaction between wording and metrics)
5.2 below.
27) Cf. Kühner-Gerth (1904:170-1).
28) In standing outside the story he tells, Homer is, in the intellectualistic terminology of
Genette (1972: 255-6) a 'heterodiegetic' narrator ('narrateur hétérodiégétique'). At the same
time he is 'extradiegetic' ('extradiégétique'): he is not, qua narrator, situated within the
framework of a larger story. Odysseus in the court of Alcinous as the narrator of his own adven
tures (books 9-12 of the Odyssey) is in two ways the opposite of 'Homer': his story falls within a
wider story and he himself plays a major role in it. Thus qua narrator he is 'intradiégétique-
homodiégétique'.
29) These terms came into use in connection with the new narrative methods and aims
practiced by Henry James. See Booth (1961: 23, 93); Genette (1972:185). The distinction be
tween 'showing' and 'telling' is useful merely to state the degree to which a narrator is 'neutral'
in the sense used here. In itself the notion of 'showing' in connection with language and litera
ture (in Aristotelian terms: mimesis) is in the last resort illusory. Language-tokens do not, by
their very nature, imitate the things to which they refer. Any act of reference, whether to ficti
cious or real-world entities, involves a deliberate choice and selection on the part of the lan
guage user. Thus when we say that Homer's aim is to 'show', rather than to 'tell', this does not
preclude that Homer imparts opinions and beliefs to his hearers. We merely say that his means
are more implicit than those of a typically 'telling' author. See Booth (1961:5-6).
30) Sometimes the Homeric narrator unexpectedly breaks this 'rule', for example in the
following passage, which occurs at a crucial point in the narrative:
Hôs phato lissomenos, mega nêpios, ê gar emellen
hoi autôi thanaton te kakon kai kêra litesthai
("So Patroclus made his appeal. But how simple he was! Had he but known it, he was pray
ing for his own doom and an evil death."), . 16,46-7. In this passage the particle ê occurs, which
is otherwise confined to direct discourse. (I owe the content of this note to prof. .J. Ruijgh.)
31) On account of the massive occurrence of direct speech Aristotle (Poetics 1448 35-36,
1460a5-ll) considered Homer to have made the best out of the epic, narrative genre, coming
nearest as possible to the best, mimetic, poetry, viz. drama.
32) Genette (1972: 203 ff.) argues that in narratology a systematic distinction should be
made between 'narration' ('qui raconte') and 'perception' ('qui voit'). The aspects of narration
pertaining to perception and point of view he calls 'focalisation'. Discussion and critique of
Genette's 'focalisation' in Bal (1978:22-9); a short exposition of focalization and an application
to Homer in De Jong (1987).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 114/318
104 LINGUISTICS AND FORMULAS IN HOMER
33) It is sometimes stated (e.g. Bennett 1981) that even in sentences like (22) is different
from even in non-comparative sentences, on the grounds that other languages (e.g. French) do
not use even (or better: its equivalent) here. But this is nonsense, of course. The distribution of
a given item in the one language cannot possibly explain the distribution of the corresponding
item in the other.
34) It is interesting to notice that scalar comparatives are in many languages expressed by
a 'continuative' propositional particle (still ', encore, nog); the comparison tends to be seen in tem
poral terms (cf. the use of already in the text). In Dutch, the difference between the scalar and
the propositional particle in scalar comparatives seems to have pragmatic significance. In 'Hij
is nog slimmer dan Paul' ('He is 'still' more clever than Paul') Paul's cleverness has been the
topic of conversation before the moment of utterance (or the addressee knows anyway that Paulis very clever), to the effect that the comparative object 'dan Paul' may be omitted ('Hij is nog
slimmer' - see ex. (26) below). In 'Hij is zelfs (nog) slimmer dan Paul' ('He is even more clever
than Paul'), on the other hand, Paul's cleverness seems to be introduced as a conversational topic
for the first time.
35) Notice, incidentally, that when the subject term is the focus constituent of even ('Even
Paul is more clever than Bill'), the sentence turns into an argument for the stupidity of Bill, in
stead of an argument for the cleverness of Paul.
36) "Ut sit maius quam alterum quod ipsum magnum est."
37) I follow Lyons (1977: 745) in distinguishing 'sentence-type' as a grammatical notion
from 'utterance'. The former denotes a property of a sentence which it has independently of ac
tual communicative circumstances, whereas the latter applies to a given speech act with a given
illocution. Thus 'declarative (sentence)' may be opposed to 'statement/assertion', 'interrogative'
to 'question', 'imperative' to 'command' and 'optative' to 'wish' The distinction is useful in that
utterances frequently have an illocution which is at variance with the sentence-type of which they
are a token. Declarative sentences, for example, may be used for persuasive purposes, see exx.
(15)-(18) in 3.2.2 above.
38) In the usual terminology and in the binary conception of modality these sentences are
treated as non-modal. See further on modality 2.1.4 above as well as 4.1.2 below.
39) Unless (27) serves as main clause to a counterfactual conditional: 'Even Paul would
have liked the excursion, if he had not been so sick'.
40) Unless the declarative sentence is used to perform an indirect speech act such as persuasion. See 3.2.2 above.
41) See Givón (1984: 321ff.). Givón points out that the referential behavior of indefinite
descriptions is altered under negation, just as in the case of other non-fact modalities (see 2.1.4):
a job in 'I don't have a job' is non-referential, while in its affirmative counterpart it cannot be
but referential.
42) Givón (1979: ch. 3; 1984:323ff.) treats negation as a separate speech act, that of denialor contradiction.
43) 'Small' forms with 'big' one of the many antonymic pairs of adjectives which are present
in any language. In such pairs one member is always 'positive' and the other 'negative'. The posi
tive member is the unmarked one in that it has the widest distribution and gives its name to the
neutral, measurable property (for example, the property denoted by the antonymic pair 'high'-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 115/318
NOTES TO CHAPTER 3 105
'low' is 'height', not *'lowness', and of the pair 'long'-'short' it is 'length', not *'shortness'). SeeGivón (1970; 1979:132-4); Hoffmann (1987:149 ff.).
44) In this example the scalar term occurs in a concessive conditional, which is introduced by kai. The example will be presented again in 63.2.1, in connection with the use of kai in concessive conditionals.
45) See Goodwin (1889:199 ff.). For the paraphrase and the conditions under which it is possible see Lehmann (1984: 330 ff.) and Bakker (1988a).
46) See also auch nur and ook maar in German and Dutch respectively (mentioned in 2.4.2above).
47) For an analysis of gar and for a comparison of its scope properties with those fo sogarand auch nur see König (1981:122-26).
48) The second part of these compounds is homos, an archaic form of the indefinite pronoun which is derived from the root sem ('one'). It functions as a negative polarity item in being solely (apart from isolated expressions like hamos ge pôs ('in some way or other')) usedin combination with oude.
49) Cf. Dik (1978:15 ff.). See also note 8 of chapter 2.
50) In Dik's Functional Grammar these terms (which are obligatory constituents, i.e. essential for the completeness of the predication) can be assigned various functions: syntactic(Subject, Object etc.), semantic (Agent, Goal) and pragmatic (Topic, Focus). See Dik (1978).
51) See Horn (1972:131 ff.), Davison (1980:11-15).52) There are some more instances of minuntha per, occurring both in wishes/commandsand elsewhere. For a discussion see 7.2.1 below.
53) The example, together with (34), is explicitly listed by Denniston (21954: 483) under'limitative' per.
54) But there are inclusive focus particles that are not scalar. The obvious example isalso/too. In other words, the scalar particles are a subgroup of the inclusive focus particles, theirspecific property being that they impose an ordering upon the alternatives for the focus constituent (viz. a scale), while normal inclusive particles do not (see König 1981:117).
55) As a possible example of an exclusive focus particle whose focus constituent represents the high point of a scale, ausgerechnet in German (uitgerekend in Dutch) may be mentioned.
In English the combination of all persons (moments etc.) is more or less equivalent.56) However, in other accounts (see Van der Auwera 1983:302; 1985:205) at least in this
use is considered to be scalar. In this approach the term 'scalartity' applies to the phenomenonof 'at least' and 'at most' readings in language. See the discussion of the various senses and usesof the word 'scale' in 2.1.1 above.
57) Denniston ( 1954:483) deals with the examples discussed in the present section partly in terms of 'limitation' and partly in terms of 'contrast'. Ex. (42), (43) as well as (33) are citedunder 'contrast', whereas per in (34) and (38) is said to be 'limitative'. This treatment not onlyfails to connect the use of per in question externally to other uses of per, it also fails to bring unityin what is a coherent and well-defined phenomenon, per (and hence scalarity) in wishes.
58) Sometimes the frequency of linguistic items in Homer is determined by the frequency of the expression(-type) in which the item occurs. For example, epic te, denoting a 'permanentfact' is disproportionately more frequent in the Iliad than in the Odyssey: of the 147 instances of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 116/318
106 LI NGUISTICS AND FORMULAS IN HOMER
hos te as a. digressive relative clause expressing a permanent fact (see Ruijgh 1971:360), no lessthan 98 occur in the Iliad and 49 in the Odyssey. But this preference of the Iliad is simply due tothe fact that the Iliad has much more epic similes than the Odyssey; 'digressive' te has a markedaffinity to timeless and, from the point of view of discourse, backgrounded, similes.
59) See Janko (1982). This study shows on the basis of quantitative research that the process of gradual linguistic innovation is in a more advanced stage in the Odyssey than in the Iliad in that older morphs are, on the whole, more frequent in the Iliad than in the Odyssey. Conversely, more recent noun endings appear to have a greater frequency in the Odyssey.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 117/318
4 PARTICIPLES I: LINGUISTICS
4.0 Introduction
We have now discussed per in connection with a number of aspects ofscalarity, viz. superlativeness, high points of scales and the intension of focusconstituents. These elements of the scalarity theory presented in the presentstudy will now be used as a background for the description of the most conspicuous and frequent use of per in Homer, per with participles. The expression-type 'participle + per 9 is such a many-sided subject that two chapters haveto be devoted to it. The present chapter deals with 'participle + per 9 from a linguistic point of view; in chapter 5 the formulaic and metrical aspects of the ex
pression-type will be discussed.The expression-type 'participle + per' provides an essential insight in the
diachronic development of per, the transition from scalarity to concession. Wesaw in 3.2.3 that relative clauses marked by per (hos per) which provide therelevant (superlative or comparative) intension of a focus constituent of kai,allow of a concessive interpretation. The notion of a scalar constituent havinga concessive meaning will form the starting-point of the discussion of 'participle
+ per 9
. It will appear in 4.2.1 that there are participles with exactly the samemeaning and function as the relative clauses of 3.2.3.But in the case of participles, 'concession' is much more important than
in the case of the scalar-concessive relative clause. We shall see that in thesemantics of 'participle + per 9 concession is an important factor in its own right,not just a by-product of scalarity. Concession is synchronically the most important aspect of the meaning of 'participle + per' in Homer. Yet scalarity (or,
better, the core of the semantics of intensional scalarity: superlativeness) is notentirely devoid of interest in the synchronic description: it allows us to makeelucidating distinctions in the material. It will appear that some of the instances of the concessive participial phrase are superlative, while others are not.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 118/318
108 LINGUISTICS AND FORMULAS IN HOMER
On this basis, I will argue that the transition from scalarity to concession is agradient rather than a discontinuous affair, some of the concessive participial
phrases being nearer to the scalar prototype than others.The discussion of 'participle + per' within the scalar-concessive
framework forms the central part of the chapter (sections 4.2 and 4.3). It is issandwiched between an introductory section (4.1), in which some of the syntactic, semantic and pragmatic properties of the participle in Ancient Greek
are discussed, and a closing section (4.4). Here I will discuss, among otherthings, the occurrence of the particle kai in the participial phrase. The sectionmay serve as a transition to the next chapter, since the distribution of kai in the
participial phrase is only to a limited extent semantically motivated. The realexplanation of the distribution lies in the metrical-formulaic sphere.
4.1 On participles
The Greek language is very fond of participles. Participles occureverywhere, in any period and in any author or text-type. As to their function
in the sentence, participles may be either 'predicative' or 'attributive'. Whenattributive, participles play a role in the determination of the reference ofnouns, just as adjectives or restrictive relative clauses.1
It is the predicative participle that constitutes the specific character ofGreek participial usage. Predicative participles may perform a number of functions. To begin, they may play a role in the complementation of certain classesof verbs. Just as attributive participles have to be discussed in connection withadjectives and relative clauses, so 'complementary' participles have to be com
pared to the second major type of subclause, the complement clause. The com plementary participle will not be discussed in the present text.3 For the
discussion of 'participle + per', only those predicative participles are of importance that may be compared to the third kind of subclause, the adverbial clause.These participles are commonly referred to as circumstantial participles
4.1.1 Kinds of adverbial relation
From a semantic point of view, circumstantial participles may be said todefine the circumstances under which the action of their verb takes place (cf.Goodwin 1889: 333). As such they have a function that may be compared tothat of adverbial subclauses. This similarity in function has to a great extent
determined the treatment of the circumstantial participle in Greek linguistics.It has been customary to state that circumstantial participles may express a
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 119/318
PARTICIPLES I: LINGUISTICS 109
number of relations (temporal, conditional, causal, concessive, etc.) in the sameway as adverbial subclauses. This treatment, however, calls for some comment.
In the case of adverbial subclauses, the nature of the relation between theaction ('state of affairs') referred to by the subclause and that referred to by themain clause is overtly expressed by the lexical meaning of the subordinator(conjunction) which introduces the subclause. A language user who wants toconvey that the relation is temporal, for example, has a temporal subordinatorat his disposal, in Greek and in many other languages even more than one.
Likewise, there are conditional, causal and concessive subordinators, and, accordingly, the concomitant types of adverbial subclause.
When using a non-sentential adverbial constituent like the circumstantial participle, on the other hand, a language user may express any given set of circumstances, but in itself the participle does not express the nature of the relation between these circumstances and the state of affairs denoted by the main
predicate. Consider, for example:
(1) a. Being rich, he spends much money on outdoor eating. b. Being poor, I would never spend money on outdoor eating.
The participles here do not by themselves express that the adverbial relationis causal or conditional, respectively. The nature of the relation has to bededuced from the context.
Being a non-sentential constituent, the circumstantial participle has nomeans to overtly express Condition, Cause, Manner, or whatever other circumstances. We may say that (Greek) participles are not meant to expressthese relations at all. The specific function of the circumstantial participle andits delimitation with regard to the function of adverbial subclauses is an important field of research, in which pragmatic approaches (the study of sentences
in discourse) are likely to be more fruitful than semantic ones. However, thisresearch falls outside the scope of the present study. What matters here is thatthe usual practice of classifying the circumstantial participle in Greek into conditional, causal, concessive, etc. is highly questionable, as it is ultimately basedupon the nature of the adverbial subclause by which we have to render the participle in a translation (see also 1.3.2 above).
The question of whether Greek participles express by themselves the adverbial relation to the main clause, has obvious implications for the study of'participle + per'. In the traditional approach, in which participles are
described as expressing the relation by themselves (and thus as virtuallysynonymous with adverbial subclauses), the function of per in participial
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 120/318
110 LINGUISTICS AND FORMULAS IN HOMER
phrases can easily be described as strengthening what is already there, viz. theconcessive force of the participle. A particularly apposite citation in thisrespect: "The participle expressing opposition, limitation or concession is oftenstrengthened by kaiper or kai" (Goodwin 1889: 341, emphasis added). In thisaccount, per is, qua strengthening element, conceived of as optional: it just mayor may not be there, there being no essential difference for the concession.10
This account can be challenged both from the point of view of per andfrom the point of view of the participle. First, when we study the adverbial rela
tion of concession in connection with scalarity, per appears to be the centralelement in the participial phrase, not just a strengthener. Second, the consequence of the account of participles presented above is that there is an essential difference between unmarked participles on the one hand and participlesthat are overtly marked for a certain adverbial relation, like 'participle + per'on the other. In the case of the former, the specific adverbial relation is a matter of context and interpretation, whereas in the case of the latter, it is overtlyexpressed. Accordingly, I describe 'participle + per' as a concessive adverbialelement which is expressed as a participle, rather than as a participle with aconcessive meaning.
4.1.2 Syntactic independence and modal autonomy
Beside adverbial semantic relations, there is a second point of view fromwhich circumstantial participles may be discussed and classified. This is the degree to which they are dependent on, or autonomous with regard to, the
predication to which they are attached. The variability of the Greek participlein this respect is the subject of a well-known monograph in Greek linguistics,Oguse (1962). In this study, the variability is discussed in terms of the modalityof the main predicate: according to Oguse, some participles may be describedas lying outside the modal sphere of the main predicate (which Oguse calls"autonomie modale"), and some participles are lying within this modal sphere("association étroite" - 'close association')11. When circumstantial participlesare 'modally autonomous', according to Oguse, they express a factual circumstance, whereas their main clause is 'modal', viz. not or not straightforwardly expressing a fact. 'Closely associated' participles, on the other hand, are justas non-factual as their main clauses. Of each kind an example:
(2) (The dying Hector to Achilles:)
êmati tôi hote ken se Paris kai Phoibos Apollonesthlon eont' olesôsin eni skaiêisi pulêisi.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 121/318
PARTICIPLES I: LINGUISTICS 111
"When your turn comes and you will be brought down at the ScaeanGate in all your glory by Paris and Apollo." (II. 22,359-60).
(3) apagou ten gunaika kai tous paidas mêden autôn katatheis."Take away your wife and your children without paying ransom forthem." (Xen. Cyr. iii, 1,37).
Ex. (2) is an instance of "autonomie modale": the addressee being esthlos('brave') is a fact which is independent of the prophesied death. The example
is cited in Oguse (1962:138). Ex. (3) is Oguse's introductory example (1962:2)of "association étroite". The participle is an integral part of the command. Theaddressee is not requested to take away his wife and children tout court; he hasto do so in this particular way. Consequently, the participle Ues in the modalsphere of the imperative verb of the main clause; the circumstance which itdenotes has not yet occurred, just as the action denoted by an imperative sentence has not yet occurred.
Now Oguse uses "modalité" in the epistemic or attitudinal sense. Whenusing 'modality' in this sense, we may oppose simple realis assertions (pure
statements of fact) to sentences which contain some modal marker (expressingvoUtion, uncertainty or doubt, for example). In this opposition the former areconceived of as entirely devoid of modality while the latter are conceived of asmodalized.13 Modahty in this sense is more or less compatible with the notionof 'modahty' in the logical tradition. It is to participles that are attached tothis kind of modalized sentence that Oguse's distinction between modallyautonomous and 'modally embedded' participles applies. Only when a sentence is modalized can we say that a participle is either autonomous withrespect to the modahty or not.
In the present study, however, the term 'modahty' is used (see 2.1.4) in a
much broader and non-binary sense, so as to comprise both the factual (realis)and the non-factual (irrealis). In this conception, the opposition betweenmodalized and unmodalized sentences is not all-important anymore, becauseevery sentence is characterized for modahty. The practical consequence of thisdifferent conception of modahty for Oguse's account of the participle is thatwhat he describes as the modal autonomy of participles turns out to be part ofa wider phenomenon. Many circumstantial participles have exactly the same
properties as Oguse's modally autonomous participles. Yet they are not recognized as modally autonomous, simply because the sentence to which they are
attached is factual, rather than 'modalized'. Consider as an example:(4) apeikhonto kerdôn, aiskhra nomizontes einai.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 122/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 123/318
PARTICIPLES I: LINGUISTICS 113
effect upon the syntactic status of the participle; it is independent as if it were'modally autonomous'.
In the argument of the present chapter I will use both notions under studyhere. As we shall see, the distinctions 'independent vs. dependent' and 'modally autonomous vs. modally embedded' are both important for the descriptionof 'participle + pef. They provide a framework within which the degree of
prototypicality (see 1.4) of instances of that expression-type may be determined. I shall argue that the prototypical instance of 'participle + pef is modal
ly autonomous (and hence syntactically independent). This property is a directconsequence of the scalar origins of 'participle + pef (see 4.3.1 below). Thetwo distinctions enable us to recognize the instances of 'participle + pef thatare either dependent or modally embedded as such and characterize them asless than prototypical instances of the expression-type in question.
4.2 From scalarity to concession
4.2.1 The intension of circumstances
In the present section I deal with the expression-type exemplified by ex.
(6) in the preliminary survey of the data-base in 1.2 above. This expression-typeis equivalent to the use of per in scalar-concessive relative clauses (see 3.2.2)in that the combination 'participle + per' is attached to a term which functionsas the focus constituent of even (kai). In the case of the scalar-concessive relative clause, this term is a proper name; in the case of 'participle + pef, on theother hand, it is another kind of extensional term, a demonstrative pronoun.
We have seen (2.1.2; 2.3.1.2; 3.2.2) that scalar superlatives and even partly determine the illocutionary potential of the sentence in which they occur:the sentence may be used for the assertion (in an indirect way) of a superlative
fact. The remarkable nature of the fact is due to a superlative property of the person whose name functions as the focus constituent of even.But the remarkable nature of a fact denoted by a sentence in which even
occurs need not always be due to a (superlative) property of one of the 'actants'in the state of affairs referred to. It may be due also to the circumstances underwhich the state of affairs in question obtained. When this is the case, the typical focus constituent of even is not one of the arguments of the predicate, buta demonstrative, anaphoric, pronominal element which points as an extensional term to a set of circumstances: even so (kai hôs in Greek).
Semantically, this demonstrative pronoun, in referring to the circumstan
ces under which the state of affairs as a whole obtains or obtained, is an ad-20
verbial element which modifies the sentence as a whole. The consequence of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 124/318
114 LINGUISTICS AND FORMULAS IN HOMERR
this is that when the adverbial is the focus constituent of even, the scope sentence of even does not coincide with the sentence referring to the state of affairs in question, as is the case in
(5) Even ACHILLES fears Hector.Scope: X fears Hector.
Instead, the scope sentence contains the nuclear predication as a whole. In aslightly formalized presentation we might say that it performs the syntactic
function Subject within the scope sentence:(6) Even SO (= under those circumstances) Achilles feared Hector.
Scope: Under circumstances X [Achilles feared Hector] obtained.
What is subject to scalarity, then, is not one of the items in the state of affairsin question, but the circumstances under which the state of affairs obtained.
Now, to the demonstrative pronoun a circumstantial participial phrasecontaining per may be attached, which defines the set of circumstances referredto which the pronoun points. The participial phrase presents the circumstances in such a way as to bring out their specific character, on account of whichthe state of affairs is something remarkable. The participial phrase provideswhat may be called in now usual terms the relevant superlative intension of thecircumstances referred to by the pronominal adverb. It explains why these circumstances have been made subject to scalarity. In other words, it explains whatis conventionally implicated by even so.
The parallelism, both in meaning and in form, between this expression-type and the expression-type discussed in 3.2.2 (per in scalar-concessive relative clauses) is obvious. In 3.2.2 the extensional focus constituent is a propername; here it is a pronominal element. And in 3.2.2 the relevant superlative in
tension is expressed by a relative clause modified by per ; here it is expressed by a participle (or participial phrase) modified hyper. As I said, the demonstrative pronoun functioning as the focus constituent of kai is hôs, which in laterGreek is usually replaced by houtôs.21 Consider now:
(7) (Nestor is boasting of the exploits of his youth: his father Neleus,thinking he was too young to join battle, hid his horses from him:)alla kai hôs hippeusi meteprepon hêmeteroisi,kai pezos per eôn, epei hôs age neikos Athênê."But even so I excelled among our horsemen, although I went on
foot. For so Athena arranged the affair." (77. 11, 721).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 125/318
PARTICIPLES I: INGUISTICS 115
(8) kai rh ' ethelen rhêksai stikhas andrôn peirêtizôn,hêi dê pleiston homïlon hora kai teukhe' arista:all' oud' hôs dunato rhëteai, mala per meneainôn.
"Hector's aim was to break the enemy line, and wherever he saw thegreatest numbers and the best-armed men he made the attempt. Buteven so he failed to break through, for the all the ferocity of his
assault." (IL 15, 617).
(9) (From the prooem of the Odyssey:)
polla d' ho g' en pontôi pathen algea hon kata thumon,arnumenos hên te psukhên kai noston hetairôn.α ll' oud' hôs hetarous errusato, hiemenos per."He suffered many hardships on the high seas in his struggles to
preserve his life and bring his comrades home. But even so he failedto save those comrades, in spite of all his efforts." (Od 1, 6).
(10) (Odysseus is being wrecked:)opse de dê rh' anedu, stomatos d' ehseptusen halmên
pikrên, hê hoi pollê apo kratos kelaruzen.α ll' oud' hôs skhediês epelêtheto, teiromenos per.
"But at last he reached the air and spat out the bitter brine that keptstreaming down his face. But even so he did not forget his boat,exhausted though he was." (Od. 5,324).
(11) (Odysseus is summoning souls from the Netherworld. The soul ofhis mother comes up:)tên men ego dakrusa idôn eleêsa te thumôi:α ll' oud' hôs eiôn proterên, pukinon per akheuôn,
haimatos asson imen, prin Teiresiao puthesthai."My eyes filled with tears when I saw her, and I was stirred tocompassion. Yet even so, deeply moved though I was, I would notallow her to approach the blood out of turn, before I had had speechwith Teiresias." (Od. 11, 88).
In each of these examples, the circumstances under which the state of affairsreferred to by the main clause obtained are anaphorically referred to as hôs('so'). What these circumstances exactly are has already been expressed inthe immediately preceding context of each of the examples. The participial
phrase in which per occurs presents the circumstances in their scalar form, justas the relative hos per -clause discussed in 3.2.2 provides the superlative proper-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 126/318
116 LINGUISTICS AND FORMULAS IN HOMER
ty of a referent, to which the scalarity is due. In both expression-types, per cannot be said to be a scalar particle proper, as neither the relative nor the participle is a genuine focus constituent. Rather, per as it occurs in (7)-(11) and inthe scalar-concessive relative clause has to be described as a marker of (inten-sional) scalarity.
The real focus constituent in the expression-type is hôs. Like all focus constituents of even (kai), it refers to what is extreme, maximally unlikely in a givensituation. Accordingly, what is meant in (7)-( 11) (viz. conventionally impUcated
by even) is that the state of affairs referred to by the nuclear predication obtained under the maximally unfavorable circumstances referred to as hôs. Forexample, the state of affairs proper in (10), the fact that Odysseus kept klinging to his raft, is made truly remarkable by the statement that it obtained underextreme, maximally unfavorable circumstances.
The participial phrase attached to hôs very naturally allows of a concessive interpretation. Just as in the case of the concessive reading of the relativeclauses in 3.2.2, the concession of the participles in (7)-(11) may be accountedfor in terms of scalarity. When we subject the circumstances under which agiven state of affairs obtains or obtained to scalarity, we thereby conventionally impUcate that those circumstances are the least likely ones under which thestate of affairs may obtain at all.23 This implies always an implicit opposition
between those unfavorable circumstances and the nuclear predication.Whenever this opposition is made visible by an explicatory constituent whichdefines the circumstances, this constituent automatically acquires the status ofa concessive adverbial element.
It follows from this scalar analysis that it I reject the kind of approach to(7)-(11) in which the participial phrase is assigned a concessive meaning whichis subsequently strengthened by kai hôs (see 4.2.2). In this approach there is,
wrongly, made no distinction between the expression-type I have just discussedand the much more frequent use of 'participle + per' as an 'ordinary' concessive phrase (see 4.3). It is important to point out that concession is, when viewedin the proper diachronic framework, the natural by-product of scalarity, ratherthan a factor in its own right.
4.2.2 Diachronic aspects of kai hôs
The saUent feature of the discussion of exx. (7)-(11) just presented is thatthe participle is described as a constituent whose function it is to explain the
presence of kai/oude and its focus constituent. Thus this discussion starts withkai hôs, describes it as a scalar expression, and proceeds from there to the par-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 127/318
PARTICIPLES I: LINGUISTICS 117
ticiple. Now this method amounts to a reversal of the usual way in which (7)-(11) and similar examples have been described in Greek linguistics. The normal way is to start with the participle, attribute to it the usual concessive 'sense'of per 24 and treat kai hôs as an adverb which strengthens the concession, just asother concessive adverbs (notably homôs and empês, see below, 4.2.3) mightdo. This practice has obviously been instigated by the fact that kai hôs, as afixed idiomatic collocation, often functions as an adverbial element expressing concession in the same way as English nevertheless, all the same or still.
In itself the attribution of the concessive value of nevertheless to kai hôs isnot unwarranted. But we have to give a diachronic dimension to this claim.Scalarity and concession are intimately related diachronically, to the effect thatconcessive meanings tend to evolve from scalar ones (see especially 4.3.1
below). Consequently, to say that kai hôs has a concessive meaning does notexclude the possibility that this meaning has once been a scalar one.
The examples (7)-(11) are best characterized as instances oîkai hôs in itsoriginal, scalar function. Kai hôs here simply means what it means decomposi-tionally: kai + hôs. There is no need to assume that kai hôs has here an
idiomatic meaning. In 4.2.3 below I will show that there is a syntactic difference between (7)-(11) and more recent examples where kai hôs does have the function of nevertheless in English.
That kai hôs in (7)-(11) has to be described in scalar, rather than in concessive terms is confirmed by the fact that (7)-(11) do not stand alone; they areintimately related to other uses of per which show an identical division of laborof per and kai, whereby kai expresses the extensional and per the intensionalside of the scalarity (see 3.2.2 and 3.3). Thus, conforming to the main thrust ofthe present study, in which we start with the notion of scalarity and proceedfrom there to uses of per that can be derived from it, we take kai hôs as a scalar
expression and proceed from it to the participial phrase and its concessivemeaning.
4.2.3 Syntactic intermezzo: correlation and apposition
When we describe the relation between kai hôs and the participial phrasein terms of 'strengthening', we have in the end to conclude that this relation is,in syntactic terms, one of correlation. 'Correlation' may be defined in syntacticterms as a relation of interdependence between a main clause and a peripheralelement (most often a subclause), in which the peripheral element is repre
sented within the main clause by a demonstrative element. Correlation is afrequent phenomenon in Greek and Latin, where the demonstrative element
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 128/318
118 LINGUISTICS AND FORMULAS IN HOMER
and the subordinator of the correlated subclause often form morphologically28
cognate pairs. Now the notion of strengthening in the present case could imply that kai
hôs as the concessive demonstrative adverbial element represents the concessive participial phrase in the main clause, whereby a relation of mutual dependence between the two would obtain. We would then have a concessivecorrelative construction in the format of 'Still/nevertheless, ...although'.However, a number of arguments can be presented against this account. First,the scalar approach advocated above is incompatible with the analysis in termsof correlation. For the syntactic consequence of the scalar approach to (7)-(11)is that the relation between kai hôs and the participial phrase has to be statedin terms of apposition, rather than in terms of correlation.
'Apposition' may be treated as a loose syntactic relation; the deletion ofthe apposed constituent has no syntactic and semantic consequences. The discourse-pragmatic motivation for apposition seems to be the expUcation of themeaning or reference of a given constituent. For instance, by way of appositionwe may identify the referent of that constituent. In exx. (7)-(11), the typical
function of the apposed participial phrase is to explain why hôs is the focus constituent of kai. In providing additional information or 'afterthoughts',30 ap
posed constituents are typically backgrounded in discourse.31
The apposed status of the participial phrases in (7)-(11) is brought out,furthermore, by the similarity, both in form and in function, of these participlesand the relative clauses (hos per) discussed in 3.2.2. The syntactic relation between a proper name and the non-restrictive relative clause attached to it can
be stated in terms of apposition;32 'correlation' has no possible applicationhere. Given this fact, it is only natural to speak about the relation between kaihôs and the participial phrase in terms of apposition, too.
The second argument is that the meaning of hôs precludes a treatment of(7)-(11) in terms of correlation. In (7)-(11), the participial phrase is invariably
placed after the main clause (the nuclear predication), and hence after kai hôs.Accordingly, when we are to assume a correlative relation between the two, bywhich the one is dependent on the other, we have to treat (kai) hôs as pointingahead to the participial phrase, i.e. as a cataphoric element. But hôs and mor
phologically cognate demonstratives are usually anaphoric, viz. referring backward to what precedes.33
This is the case, too, in (7)-(11). The surprising circumstances under which
the state of affairs obtains, are always described in the immediately precedingcontext, and hôs refers back to them. When this fact is not recognized, we do
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 129/318
PARTICIPLES I: LINGUISTICS 119
no justice to the discourse function of kai hôs in (7)-(11). The participial phrase,in its turn, in presenting the circumstances in their scalar form, refers back tohôs. Matters would be entirely different when the participial phrase precededthe nuclear predication and kai hôs. Then the case for correlation would bemuch stronger, hôs referring back to the participle, and the participle, conversely, pointing ahead to hôs.
Concessive correlative constructions in which 'participle + per' precedesthe nuclear predication, do not occur in Homer. They only appear in later times,
when 'participle + per' has long become a modifier which is in itself concessive, rather than scalar (see 4.3.1). Below, two examples of the post-Homericconcessive correlation are presented. In (12) the participial phrase (in its final,
post-Homeric form: 'kai per + participle') uncharacteristically (from the pointof view of Homeric usage) precedes the nuclear predication. Instead of kai hôs,the nuclear predication contains the concessive adverbial homos, with whichthe participle is interrelated, homôs pointing backwards to the participle, andthe latter pointing forwards to homôs:
(12) hoi de hate Perìandrou eonta paida, kaiper deimainontes, hornos
edekonto."And because he was the son of Periander, they still received him,although they were scared." (Hdt. iii, 51,3).
In (13) below the participle is not modified by per, but it equally precedes thenuclear predication. The latter does contain kai (oud') hôs. But here the adver
bial has its final concessive function, which differs both semantically and syntactically from the original scalar function of which (7)-(11) above areexamples:
(13) es to auto ksunelthontes, oud' hôs edunêthêsan ksumbênai."And though they met together again afterwards, they nevertheless
did not come nearer to reaching an agreement." (Thuc. v, 55,2).
But with these examples we are long past the transition from scalarity to concession. This applies to per in (12) as well as to oud' hôs in (13). Both the participial phrase (whether or not modified by per) and the adverbial element(whether homôs or kai hôs) have a function which is altogether differrent fromwhat we have seen in (7)-(11)36
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 130/318
120 LINGUISTICS AND FORMULAS IN HOMER4.3 The concessive participial phrase
We now arrive at the discussion of the use of 'participle + per*, as an ordinary concessive, the single most conspicuous and frequent use of per inHomer.37 Concession has here become a factor in its own right. But scalarityis still an important concept. It gives relief to the mass of instances, in enablingus to draw distinctions in the material which remain otherwise unperceived. Torecognize the scalar origin of 'participle + per' is to have an open eye for thevariability which is present in the material, both in diachronic and in synchronic
space. And to recognize this variability is necessary for the description of theexpression-type in question.
The insight that scalarity is the origin of concession, or, alternatively, thatconcession is the 'product' of scalarity, greatly favors the unified description of
per. In 1.3.2 I argued that to deal with 'concession' as a possible sense of per isunsatisfactory in that the relation between this sense and other senses remainsin the dark. But when we conceive of 'concession' as one pole of the axis'scalarity-concession', the relation between the concessive and the non-concessive is can be better accounted for. Notice, incidentally, that the easy transition
(in diachronic space) and close connection (in synchronic space) is a phenomenon which is not confined to Ancient Greek. In Dutch, for instance,the concessive subordinator hoewel must have a scalar origin ('hoe-wel' -'however much').
4.3.1 The prototypical properties of 'participle + per'
The dividing-line between 'participle + per 3 as used in exx. (7)-(11), viz.as an apposed scalar phrase, and the use of 'participle + per' as a concessive isnot a clear-cut one. The instances of 'participle + per' differ from one anotheras to the degree to which they have the properties of the scalar participial
phrase. Some instances of 'participle + per' are definitely nearer to their scalar'roots' than others. In other words, we may speak in terms of a continuum inwhich instances of 'participle + per' can be ordered by degree of decreasingaffinity with scalarity. This means that the scalar(-concessive) participle exemplified by (7)-(11) may be conceived of as the prototype (see 1.4), and allthe concessive participles as less than prototypical, in various degrees.39
The prototypical features of the scalar participial phrase, then, on the basisof which the instances of 'participle + per' can be classified and localized inthe continuum, may be described as follows.
1) Superlativeness. The prototypical scalar participle denotes somethingsuperlative, since the very point of asserting sentences like (7)-(11) above was
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 131/318
PARTICIPLES I: LINGUISTICS 121
that the circumstances under which the state of affairs in question obtained (ordid not obtain) were maximally (un)favorable. The introduction of'superlativeness' as a factor in the description leads to a useful differentiationin the material. When using the concept we perceive that in the case of analytic
participial phrases (type 'predicate + per + eon ('being')'), there is a difference between adjectival and substantival predicates. Analytic participial
phrases which contain a substantival predicate are less prototypical as regardsscalarity than analytic participles with an adjectival predicate, because super-
lativeness cannot apply to them (see further 4.3.2.2). But not all analytic participles with an adjectival predicate are prototypical, because not all adjectivesdenote properties which allow of degree and comparison (see 4.3.2.1).
In the case of synthetic participles, the introduction of superlativeness asa Leitmotiv makes us perceive that participles denoting state-like, time-stable
properties (such as akhnumenos ('grieved'), hiemenos, memoôs ('eager')) haveto be distinguished from participles denoting more time-unstable, event-like(kinetic) phenomena (such as pinôn ('drinking')). The latter are clearly less
prototypical as regards scalarity than the former, again, because they do notallow of superlativeness. See further 4.3.2.3.
2) Independence. This second prototypical property is a consequence ofthe explicatory function of the participles in (7)-(11). Being an apposed element, its deletion does not affect the syntactic integrity of the sentence to whichit is apposed. Participial phrases with per can only be dependent (i.e. contributing essentially to the completeness of their main clause, see 4.1.2) when theyare disconnected from scalarity. See further 4.3.3.
3) Modal autonomy. The prototypical (scalar) participial phrase is notonly independent, but also 'modally autonomous' in the sense of Oguse (1962).It stands outside the scope of the propositional modalities of the main clause:
the prototypical instance of 'participle + pef is factual, irrespectively of itsmain clause being factual or non-factual. The specific scalar variant of 'modalautonomy' which we find in the case of 'participle + pef is due to the semantics of even. We saw above (2.4.1) that even has to be analyzed as a positive
polarity item which refuses to stand inside the scope of negation (except whenit occurs in a negative polar context, see 2.4.2 and 3.4.3). The 'autonomy' ofeven with regard to negation (as well as to other sentential modalities) is im
parted to what is related to its focus constituent by way of apposition. Thecriterium of modal autonomy enables us to recognize those instances of 'participle + pef that are 'modally embedded', and to attribute to them a less than
prototypical status.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 132/318
122 LINGUISTICS AND FORMULAS IN HOMER
We now go on to discuss 'participle + per' within the framework of thesethree prototypical features. The examples cited below possess the three properties; accordingly, they are minimally removed from scalarity:
(14) (Odysseus has just given the braggart Thersites a good beating:)hoi de kai akhnumenoi per ep' autôi hêdu gelassan."The rest, disgruntled though they were, had a hearty laugh at hisexpense." (7Z. 2,270).
(15) (Hector, who is away from the battle for a while, to Helen:)mê me kathiz', Helenê, phileousa per, oude me peiseis."Helen, you are kind, but do not ask me to sit down [however much
you care for me...]." ( . 6,360).
(16) (Poseidon urges the Greeks:)iomen: autor egôn hêgêsomai, oud' eti phêmi
Hektora Priamidên meneein mala per memaôta
"I myself will take command, and I do not think Prince Hector willstand up to us long, for all his fury." (7Z. 14,375).
(17) (Antilochus joins Menelaus in his fight with Aeneas:) Aineias d'ou meine, thoos per eôn polemistês,
hôs eiden duo phôte par' allêloisi menonte."And when Aeneas saw the two men making this united stand, hefelt unable to face them, for all the daring he had shown before."
(IL 5, 571).
(18) (Amphion and Zethus:)hoi prôtoi Thêbês hedos ektisan heptapuloio
purgôsan t' epei ou men apurgôton g' edunantonaiemen eurukhoron Thêbên, kraterô per eonte.
"The founders of Thebes of the Seven Gates, who first fortified itssite with towers, since for all their prowess they could not establishthemselves in the open lands of Thebes without a wall to their city."(Od 11,265).
(19) (Hephaestus about his wife Aphrodite and Ares, who have just beencaught in bed together:)ou men spheas et' eolpa minuntha ge keiemen houtô,
kai mala per phileonte.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 133/318
PARTICIPLES I: LINGUISTICS 123
"Yet I have an idea that they won't be eager to prolong that embrace,no, not for a moment, not for all their love." (Od. 8,316).
The participial phrases in these examples possess all the three prototypical properties discussed above. Ah of them denote a state or property that allows of degree and comparison. Hence superlativeness is involved. Secondly,all of (14)-(19) are independent, their main clause being complete withoutthem. In (17), for example, the bare main clause 'Aeneas did not stay' is in itself a meaningful contribution to the flow of narration; it does not need the participial phrase to achieve that status. Thirdly, the participles are not affected
by the modality involved. In (16), (17), (18) and (19), this modality is a simplenegation. In (15) we have a negative command; the participle stands outsidethe scope of both the negation and the command: 'Don't...., (and I ask you this)in spite of...'.42
Yet exx. (14)-(19) are one step removed from scalarity: the absence ofkailoud' hôs makes a considerable difference. In the discussion of exx. (7)-(11),I stated that the point in these examples is the assertion not so much of the stateof affairs envisaged in the nuclear predication as of the fact that this state of af
fairs obtained under maximally unfavorable circumstances (or, conversely, didnot obtain under maximally favorable circumstances). Thus, conforming to thesemantics of even (see 2.3.1.2), exx. (7)-(11) are normal statements of superlative facts.
The very point of uttering (7)-(11) is the statemment that the nuclear predication obtained under the circumstances referred to as kai hôs. This is thesuperlative fact. In (14)-(19), on the other hand, kai hôs is absent. Consequently, these examples have a different illocution and discourse function. They arenot uttered as an argument for some superlative fact. Rather, we have to saythat the nuclear predication here is uttered for its own sake. The participial
phrase has been added as a normal concessive adverbial constituent, ratherthan as an apposed constituent with an explicatory function. Of course, the participial phrase lends salience to the nuclear predication. In (14), for example,the Greeks' laughing while they are akhnumenoi is more remarkable than whenthey are not. But this is the normal salience conveyed to a main clause by anyconcessive adverbial constituent. The salience does not make of the nuclear
predication the statement of a superlative fact. Thus we may say that exx. (7)-(11) are related to (14)-(19) in the same way as 'Even Paul liked the excursion'is related to the simple Paul liked the excursion'. The former functions as an
argument for a superlative fact; the latter does not, irrespectively of its beingsalient or not.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 134/318
124 LINGUISTICS AND FORMULAS IN HOMER
Notice that when we insert kailoud' hôs in (14)-(19), there is nothing inthe preceding discourse to which hôs as the focus constituent of kai could referanaphorically. If kailoud' hôs occurred in examples like (14)-(19), its only possible function would be to point ahead to the participial phrase. But kai hôsdoes not occur in examples like (14)-(19), and this corroborates the arguments
put forward in 4.2.3 above: hôs cannot be cataphoric (i.e. pointing ahead to the participle). If it could, it would sometimes occur in examples like (14)-(19). Butit never does.
We now go on to show that the 'process' of becoming an ordinary concessive extends beyond the kind of instance exemplified by (14)-(19). In otherwords, we are going to proceed from the prototypical instances of 'participle+ pef to the less than prototypical ones. These instances have only one or twoof the prototypical properties mentioned above, not all the three of them. Mostcommon are those examples that lack the first property, superlativeness. Thenon-superlative participial phrase is disconnected from scalarity in an essential respect, but nothing precludes its use as a concessive adverbial element.
Instances that lack the second and third prototypical properties, on theother hand, are much less common. This is not surprising when we realize thatindependence and modal autonomy are prototypical properties of concessivesin general, irrespectively of their having a scalar origin. Thus, in contrast withthe loss of superlativeness, the lack of either of the other two properties is atvariance with the central synchronic meaning of 'participle + per as a concessive. In other words, dependent and/or modally embedded instances of 'participle + pef (which will be discussed in 4.3.3 and 4.3.4) are less than
prototypical anyway, whether we conceive of the prototype as scalar (involvingsuperlativeness) or not.
Consequently, the prototypical properties 2) and 3) differ from 1) in that
they apply both in the diachronic and in the synchronic dimension. Independent and/or modally embedded instances of 'participle + per' are peripheral both from a synchronic and from a diachronic point of view. Prototypical property 1), on the other hand, exclusively applies in the diachronic dimension,since synchronically, a non-superlative instance can make a perfect (prototypical) concessive phrase.
4.3.2 Loss of superlativeness
Superlativeness is the key semantic concept in intensional scalarity. We
might say that the superlativeness of an instance of 'participle + per' is a necessary condition for its being a scalar expression: if it is not superlative, it cannot
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 135/318
PARTICIPLES I: LINGUISTICS 125 be scalar either. Now, superlativeness puts lexical and categorial constraints onthe participial phrase. If the participial phrase is analytic (type 'predicate + per+ eon9), the predicate has to be (i) an adjective, substantives being qua categoryexcluded from superlativeness, and (ii) an adjective that may allow of superlative marking at all. In other words, it must be an adjective which denotes a
property that may be measured and compared. It is this type of adjective, ofcourse, that may also appear as a scalar superlative in a non-factual environment. For example, the participial phrase in (18)(kraterô per eonte) contains
an adjective (krateros, 'strong') which may appear in the superlative mode, andwhich, accordingly, may function as a scalar superlative, see for example ex. (4)in 3.1.1 above.
If, on the other hand, the participle is synthetic, it has to possess the features of real adjectives; it has to denote a property rather than an event. Goodexamples of participial phrases with such a participle are akhnumenos,hiemenos, teiromenos per) ('grieved, eager, exhausted as/though he was'), aswell as mala per meneainônlmemaôta ('furious as/though he was') and
phileousa per ('however much she loved').
Now given this characterization of the superlative participial phrase, theloss/lack of superlativeness opens up a number of possibilities. In the case ofanalytic participles, the predicate is either an adjective that does not allow ofsuperlativeness or it is a substantive, and in the case of synthetic participles, the
participle denotes a more time-unstable, kinetic entity than the 'adjectival' participle. These possibilities will now be dealt with in the following subsections.
4.3.2.1 Non-superlative adjectives
It is possible to construct a continuous space in which predicates are lo
calized by degree of the time-stability of the entity which they denote. On theone extreme of this continuum, we have verbs which denote phenomena witha minimum of time-stability, viz. events. On the other extreme we have substantival, category-ascribing predicates with a maximum of time-stability. Qua
predicates, adjectives are situated somewhere in between.44 The propertiesdenoted by adjectives are situated midway on the time-stability scale. Now it is
precisely these properties that may be measured and compared. The possibilityof measurement is reflected in language by the morphosyntax of the 'degreesof comparison'.
However, not all adjectives can appear in the superlative mode. This ap
plies in particular to adjectives that verge on the substantival, for example, ad jectives denoting materials. Something either is made of gold, or it is not; there
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 136/318
126 LINGUISTICS AND FORMULAS IN HOMER
are no degrees possible to which this is the case. And yet this kind of adjectivemay be combined with per in a participial phrase. The resulting concessive isdisconnected from scalarity:
(20) (From the description of the new shield for Achilles. A ploughingscene is depicted:)hê de melainet' opisthen, arêromenêi de eôikei,khruseiê per eousa."The field, though it was made of gold, grew black behind them, asa field does when it is being ploughed." (II. 18,549).
Similar is the following example, though lithos ('stone') is properly speaking asubstantive:
(21) (Niobe has been punished by the gods. She stands petrified in thehills of Siphylus:)entha, lithos per eousa, theôn ek kêdea pessei."There Niobe, though in marble, broods on the desolation that thegods dealt out to her." (II. 24,617).
Other adjectives are equally near-substantival. In Greek there is no formal difference between 'foreign' and 'a foreigner':
(22) (The Trojans hear the news that Sarpedon, the king of the Lyciansis dead:)Trôos de kata krêthen labe penthosaskheton, ouk epideikton, epei sphisin herma poleoseske, kai allodapos per eoa
"The Trojans were heart-broken at the news: it was intolerable.
Sarpedon, though a foreigner, had been a butress of their city." (II..16,550).
In other cases, the non-superlativeness is due to other factors than the(near)-substantival status of the adjective and the time-stability of the property which it denotes. For example, 'being on one's own' is not a particularly time-stable affair. Yet it does not allow of measurement. A warrior is or is not on hisown; there is no room for degree. So a participial phrase containing this attribute cannot be but purely concessive:
(23) Tudeidês d' autos per eôn promakhoisin emikhthê.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 137/318
PARTICIPLES I: LINGUISTICS 127
"Although he was left to his own resources, Diomedes drove up tothe point of attack." (R. 8,99).
In the following example, the absence of superlativeness is due, paradoxically,to the fact that the adjective appears in the superlative mode:
(24) (A comment on Hephaestus having caught Ares in a net while thelatter is making love to Aphrodite:)ouk aretâi kaka erga: kikhanei toi bradus ôkun,
hos kai nun Hêphaistos eôn bradus heilen Areaôkutaton per eonta theôn hoi Olumpon ekhousii"Bad deeds don't prosper. The tortoise catches up the hare. See howour slow-moving Hephaestus has caught Ares, though he is the
fastest of the gods who dwell on Olympus." (Od. 8,331).
'Fast' can be measured, but 'fastest' cannot: you are either the fastest or youare not. There is no room for degree.
The following example is similar to (24) in that the participial phrase contains an adjective that in itself can be perfectly subjected to superlativeness. In
(24) it is superlativeness that neutralizes the superlativeness; in (25) it is negation:
(25) (The embassy to Achilles has failed. Aias to Odysseus just beforethey leave:)apaggeilai de takhistakhrê muthon Danaoisi kai ouk agathon per eonta
"We must at once report the news to the Danaans, though it is notgood news." (R 9,627).
A negation in a per-phrase, both in Homer and in later Greek, is a very rarething. The negation resolutely destroys the superlative potential of agathos,for the absence of a property is absolute: it cannot be subjected to comparisonand superlativeness. The negation in (25) straightforwardly shows that this example belongs to the group of instances in which the connection with scalarityhas disappeared, and where per has lost its original meaning. The participial
phrase has become an ordinary concessive.Sometimes a synthetic participle is equivalent to a non-superlative adjec
tive. The best examples are perfect participles meaning 'dead' or 'bornout/from'. Consider:
(26) (Hector offers a reward in the struggle over Patroclus' body:)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 138/318
128 LINGUISTICS AND FORMULAS IN HOMER
hos de ke Patroklon kai tethnêôta per empêsTrôas es hippodamous erusêi, eiksêi de hoi Aias,hêmisu toi enarôn apodassomai, hêmisu d'autos."To the man who forces Aias back and brings Patroclus, dead though
he is, into the Trojan lines, I will give half the spoils, keeping half formyself." (IL 17,231).
(27) (Achilles to Asteropaeus, whom he has just slain:)keis'houtô: khalepon toi eristheneos Kroniônos
paisin erìzemenai, potamoio per ekgegaôti."Lie there, and learn how difficult it is for you, even though you area child of a River-god, to fight the scions of almighty Zeus." ( . 21,185).
4.3.2.2 Substantival predicates
It is clear that what appUes to adjectives verging on the substantival ap plies to substantives themselves too. Substantival predicates are, all otherthings being equal, non-superlative, because the entities they denote are too
time-stable to allow of measurement.In the scalar framework within which per has been discussed in chapter 3,
we may predict that the combination of per with a substantive, viz. a term whichis not (purely) intensional, does not normally occur. Yet in exx. (28)-(30) thecombination does occur, as a consequence of the fact that the participial phraseis simply concessive, rather than scalar.
(28) (Eurylochus has accused Odysseus of risking his comrades' life forthe sake of his own recklessness:)hôs ephat', autor ego ge meta phresi mermêrìksa,spassamenos tanuêkes aor pakheos para mêrou,toi hoi apotmêksas kephalên oudasde pelassai,kai pêôi per eonti"Now when Eurylochus said that, I had half a mind, though he wasa close kinsman of my own, to draw the long sword from my side andlop his head off to roll in the dust." (Od. 10,441).
(29) (An encouragement of Agamemnon:)Teukre, philê kephalê, Telamônie, koirane laôn,ball' houtôs ai ken ti phoôs Danaoisi genêai
patri te soi Telamoni, hos s'etrephe tutthon eonta,
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 139/318
PARTICIPLES I: LINGUISTICS 129kai se nothon per eonta komissato hôi eni oikôi."Teucer, son of Telamon, my beloved prince, shoot as you are doingnow, and you may well bring salvation to the Danaans and fame toyour father Telamon, who took you under his roof and reared youthough a bastard child." (II. 8,284).
(30) (Hermes reassures Priam: his son's body is still in Achilles' tent andit is not in decay:)
hôs toi kêdontai makares theoi huios heêoskai nekuos per eontos, epei sphi philos peri kêri."Which shows what pains the blessed gods are taking in your son's
behalf though he is nothing but a corpse, because they love himdearly." (II. 24,423)47
In these examples, the substantive cannot be described as the (extensional)focus constituent of the scalar particle per. Rather, we have to say that per, asa kind of propositional particle, marks as concessive a participial phrase ofwhich a substantive is the predicate.
4.3.2.3 Adjectival and verbal participles
In the two previous subsections we saw that when we construct a continuum of time-stability of predicates, and move from the middle towards thearea of more time-stability, where the substantival predicates have to be located, we reach a point where superlativeness is not possible anymore. Now,the same happens when we move in the opposite direction. Typically time-unstable phenomena like actions and events do not allow of degree and com
parison. Consequently, the morphosyntax of the 'degrees of comparison' does
not apply to the verbal word, just as it does not apply to the category on theother extreme of the scale, the substantive. It is with this phenomenon that the present subsection is concerned.
Nominal (i.e. adjectival or substantival) predicates differ from verbal onesin that they are aspectless. This applies to the analytic participial phrase too,obviously because it contains a nominal predicate. The typical aspectless statement is the ascriptive (or attributive) sentence (see Lyons (1977: 185,417)) ofthe general form 'S(ubject) is P(redicate)'. In such a sentence, one eitherascribes a property to a subject, or one assigns a subject to a given category.Any analytic participial phrase may be reduced to such a sentence, whereby the
participial form of 'be' functions as the copula. The participial phrase in (29),for instance, is attached to one of the arguments of the nuclear predication in
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 140/318
130 LINGUISTICS AND FORMULAS IN HOMERsuch a way as to reflect a simple subject-predicate expression: 'Teucer is a bastard (child)'.
Synthetic participles, too, may be aspectless. The aspectless synthetic participle with adjectival properties yields many instances of 'participle + per':akhnumenos, hiemenos, teiromenos per, mala per meneainôn/memaôta).These participles belong without exception to the superlative kind and as suchthey are prototypical with regard to scalarity.
However, not all synthetic participles are of the adjectival, state-denoting
kind. In fact, most synthetic participles denote phenomena with considerablyless time-stability than akhnumenos and meneainôn. These participles tend toreport, in the manner of finite verbs, what a subject is doing in a particular situation, while participles like akhnumenos and meneainôn tell what a subject is.
Activity/event-denoting participles are definitely non-adjectival in nature.In the case of these participles, there is no connection with simple ascriptivesentences in which a copula is involved. They are characterized for two of themorphosyntactic categories of the verbal word, aspect (fluctuation betweenaorist and present forms) and 'diathesis' (active, middle and passive forms).These morphosyntactic possibilities are a direct reflection of the nature of the phenomena denoted. Unlike states, activities and events may be presented asongoing or completed, and this is encoded in the language by the morphosyntactic opposition between aorist and present. Secondly, actions may bereported in discourse as something done by an agent, or as something undergone by a patient. Here the morphosyntax of 'diathesis' has its role.
Once the morphosyntax of aspect and diathesis has entered the picture,we have left the area on the time-stability scale where superlativeness is a factor. And this has, of course, a clear ontological basis: 'what you are' can bemeasured; it is a matter of degree. 'What you are doing', on the other hand, is
absolute: you either do something or you do it not.52
It follows from this thatwhen per is attached to an activity-denoting participle, we get a concessive participial phrase which is entirely dissociated from the scalar origins of 'participle+ per', superlativeness being neutralized. There are very few instances of 'activity-denoting participle + per'; I have found only three. Two of these are
presented below; the third one will be discussed in 4.3.4 below, in connectionwith 'modal embeddedness'.
(31) (Aias to Teucer in a very difficult situation:)mê man aspoudei ge damassamenoi per heloien
neos eusselmous.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 141/318
PARTICIPLES I: LINGUISTICS 131
"I bet it will not be without an effort that the Trojans will capturethe well-found ships, though they have beaten us already." (II. 15,476).
(32) Nestora d' ouk elathen iakhê pinonta per empês."The din did not escape the ears of Nestor, though he sat drinkingin his tent."(II. 14,1).
In (31) the participle is aorist and in (32) present. In (31) the action denoted
by the participle is completed - and this is an essential part of the meaning ofthe participle - whereas in (32) the action is presented as ongoing. In neithercase superlativeness is a possible factor.
4.3.3 Dependent participles
In 4.3.1 I stated that in its original use as an apposed constituent, 'participle + pef is independent (prototypical property no 2). 'participle + per' isa typically 'backgrounded' constituent; it does not contribute essentially to thesentence to which it is attached. This property of the scalar participial phrase
is retained in the transition from 'scalar' to 'concessive' to a far higher degreethan superlativeness. 'Independence' remains an important feature of the concessive participial phrase that is the result of the gradual fading of the scalarityof per. The reason is, of course, that syntactic independence may be considereda property of the concessives in general, whether they have a scalar origin ornot.
I have found only one instance of 'participle + pef functioning as a de pendent participial phrase. Consider:
(33) (Achilles prays to his mother Thetis:)
meter, epei m' etekes ge minunthadion per eonta,timen per moi opheïlen Olumpios eggualïksai
Zeus hupsibremetês, nun d' oude me tutthon eteisen."Mother, since you gave me life, if only for a little while, surelyOlympian Zeus the Thunderer owes me at least honor. But as it ishe pays me none!" (7Z. 1,352).
The second per in this passage (timên per) has already been discussed; it is aninstance of scalarity in wishes (see 3.5.3 above). We are now dealing with thefirst one. The participial phrase in (33) is one of the better known in Homer;
it has repeatedly been mentioned separately. Kühner-Gerth (1904:169) seemto suggest that per here means 'very' ('sehr'). This would imply that the par-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 142/318
132 LINGUISTICS AND FORMULAS IN HOMER
ticipial phrase in (33) is not properly concessive but an instance of what Den-niston calls the 'intensive use', the alleged original intensifying use of per (see1.3.4 above). However, this treatment is unwarranted, because (i) 'intensifying
pef is an unnecessary concept in the description, and (ii) the example can (hasto) be accounted for otherwise.
The participle must be assigned its usual concessive function, as Dennis-ton (21954:482) rightly remarks. But the example is not less remarkable at that.The concession in (33), and hence the proper meaning of the participial phrase,
must be in the opposition between being a goddess' son and being short-lived.If the participle had been attached to a normal main clause, there would benothing strange: the participial phrase would have been a normal independentone (Ί am a goddess' son, short-Uved though I am'). But the sentence to whichthe participial phrase is attached is not uttered for its own sake; being a causalsubordinate clause, it leads up to ('sets the scene' for) the following discourse.And it is with respect to the following discourse that minunthadion per is anecessary constituent, without which the causal subordinate clause is incom
plete.
After saying 'You gave me life' to his mother, Achilles goes on to utter
what was called above (3.5.3) a 'compromise wish'. Here the second instanceof per {timên per) plays a major role. The highest item on Achilles' desirabilityscale (an additive one, see 3.5.3) is honor plus longevity. But since longevity has
been denied him, he has to content himself with a compromise: honor alone,the second best item on the desirability scale and the least item that still satisfies his wish.
The participial phrase of the preceding subclause is indispensable in thecontext of (33), because in telling that Achilles is short-Uved, it provides thevery reason why Achilles has to resort to the compromise at all. Achilles does
not want to say 'Zeus has to grant me at least honor since you gave me life'. His point is 'Zeus has to grant me at least honor since I am short-lived'. In a highly unusual way (from the point of view of Homeric per, that is), the participial
phrase is the most salient item in the causal subclause. It contributes essentially to the realization of Achilles' communicative intentions.
The concessive function which the participial phrase has qua instance of'participle + pef is strictly speaking unnecessary in the discourse. But thenAchilles is not deHvering a tightly organized speech. At first, minunthadionperseems to be in the usual way concessive with respect to what precedes. But whenthe discourse goes on, the participle turns out to have a causal function with
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 143/318
PARTICIPLES I: LINGUISTICS 133respect to what follows. On account of the causal use to which it is eventually
put, the participial phrase is clearly a less than prototypical instance of its kind.
4.3.4 Modally embedded participles
We now arrive at the discussion of the third prototypical property mentioned above (4.3.1), modal autonomy. I have argued that the prototypicalscalar instance of 'participle + per' stands outside the scope of the sententialmodalities of the predication to which it is attached. This has to be viewed in
connection with the semantics of even, as we have seen. But 'modal autonomy'applies to the majority of the concessive participial phrases as well. There are,however, some significant exceptions. Consider first:
(34) (Athena is abusing Ares for wanting to fight. If he does it, Zeus will be very angry:)ê etheleis autos men anaplesas kaka pollaaps' imen Oulumponde kai akhnumenos per anankêi,autor toh alloisi kakon mega posi phuteusai;"Or do you wish to get a thrashing for yourself and to be chased backto Olympus, grieved as you are, while the rest of us reap thewhirlwind you have sown?" (77. 15,133).
The participial phrase here is attached to an infinitival clause which is the com plement of the verb 'want' (etheleis). The complement of this verb alwayscreates a non-factual environment (see 2.1.4). The participle is under the scopeof this modality: the overall construction of (34) is not 'Or do you want, grievedas you are, that...', but 'Or do you want that grieved as you are ...'. Ares' beingakhnumenos is not an independent fact; it depends on what he wants. In beingwhat I call 'modally embedded', rather than modally autonomous, the par
ticipial phrase in (34) is in a conspicuous way a less than prototypical instanceof its kind.55
The following three instances have received relatively much attention inthe accounts of per. This is understandable, for the modally embedded statusof the participle makes the concession a great deal less conspicuous than wemight want or expect. Consider:
(35) (Agamemnon, when Achilles has said that he (Ag.) will receiveample compensation for the loss of Chryseis:)mê dê houtôs, agathos per eôn, theoeikel' Akhilleu,klepte noôi, epei ou pareleuseai oude me peiseis.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 144/318
134 LINGUISTICS AND FORMULAS IN HOMERR
"You are a great man, Achilles, but do not imagine you can trick meinto that. I am not going to be outwitted or cajoled by you." ( . 1,131).
(36) mê m'eti nun kateruke, lilaiomenon per hodoio
"I am eager to be on my way; please do not detain me now." (Od. 1,315).
(37) (Telemachus to his mother Penelope:)
meter emê, mê moi goon ornuthi mêde moi êtoren stêthessin orine phugonti per aipun olethron.
"Mother, please do not reduce me to tears or play on my emotionswhen I have just escaped such a deadly fate." (Od. 17,47).
These examples have been dealt with in three different ways, two of which arethe consequence of the fact that the concession of 'participle + pef here is notas straightforward as one might expect. The first method to explain the allegedabsence of concession in (35)-(37) is the approach which crops up from timeto time in the study of per, the 'original meaning-approach' (see 1.3.4), which
has also been applied to (33) above. One of the commentators on II. 1,131 ( =ex. (35)): "Per seems here to have merely its original force of 'very', rather thanof 'though', which indeed belongs properly to the participle." (Leaf 1900-2 i,14). But once more the assignment of an instance of per to the 'original''intensive' use turns out to be unnecessary and unwarranted. The participial
phrases can perfectly be accounted for in the synchronic dimension. Besides,how are we to account for 'intensification' in the case of (37)? The participlehere denotes an event, rather than a property. Hence 'intensification' cannot
possibly apply to it (see 4.3.2.3 above).
The second approach is practiced in an older Homeric lexicon (Ebeling1885: ii 162B s.v. per). It is stated here that the participial phrase in (35)-(37)is causal, rather than concessive. Indeed, when we stick (consciously or not)to the independent and modally autonomous status of all the instances of 'participle + pef, and interpret the participles in (35)-(37) in the same way as
phileousa per in (15) above, it is hard, if not impossible, to conceive of the participles as concessive. In the case of (35), for example, the paraphrase which
brings out the alleged independent and modally autonomous status of the participle ('Don't cheat, Achilles, and I ask you this although you are a great man')does not make much sense: we expect something Hke because instead of al
though. Likewise, it is more natural to read (36) as 'Do not detain me nowbecause I am eager to be on my way'.58
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 145/318
PARTICIPLES I: LINGUISTICS135
But being forced to say that the participial phrases in (35)-(37) are causalrather than concessive is a very high price to pay for the account of it as a normal independent and modally autonomous participle. To describe the participles in (35)-(37) as causal is ruinous for the unifying description of per, forhow are we to account for causal adverbial constituents in the scalar-concessive framework? If the unifying, scalar-concessive framework has any value, wehave to reject the causal interpretation. The best policy, then, is to stick, conforming to the principle of unification (see 1.3.2), to concession as long as pos
sible, and accept the consequences of that decision. These consequencesinvolve giving up the modally autonomous status, and hence the scalar connection of the participles. But then modal autonomy is merely a prototypical
property of 'participle + per', whereas giving up concession is to deny its veryessence.
The third approach to (35)-(37), advocated by Denniston,59 recognizesthe modal embeddedness of the participial phrase. It is, basically, the right one.Discussing 'participle + per' in connection with negation, Denniston states:"Either the participial clause is contrasted with the negation of the idea (...), orit is contrasted with the idea which is negatived, as regarded by itself positively." (Denniston 1951: 485). It is not difficult to recognize in the first alternative the independent, modally autonomous instances of 'participle + per' thatstand outside the scope of negation. The second alternative is a cumbrous wayof saying that 'participle + per' may be under the scope of negation. Dennis-ton gives (35)-(37) as examples of his second alternative.
Above (4.1.2 and 4.3.1) I characterized a modally autonomous participleas a participle that is factual irrespectively of the modality of the main predicate. By contrast, a modally embedded participle would then be non-factual,as it is in the modal sphere of a non-factual main predicate. However, matters
are not that simple. The participle in ex (3) of section 4.1.2 above, as well asakhnumenos per in (34) are non-factual in the way just mentioned. But the participles in (35)-(37) are not. In none of these examples is there any doubt as tothe factuality of the participial phrase. But this does not detract anything fromtheir modally embedded status. The point here is that the predicate to whichthe participles in (35)-(37) are attached is factual, in spite of its appearing inthe imperative mood.
Imperative predicates may denote a factual state of affairs, namely whensomeone is requested to stop doing something. Ancient Greek has a charac
teristic way of expressing such commands: the prohibitive-negative particle mê plus a present imperative. Commands of this type are feUcitous speech-acts
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 146/318
136 LINGUISTICS AND FORMULAS IN HOMER
either when the addressee is about to perform the action in question, or whenhe is performing it already. In the latter case the action is factual on the moment of utterance.60 This is precisely what happens in (35)-(37): the negative
present imperative is used to request the addressee to stop cheating thespeaker, detaining him and working on his emotions, respectively.
To recognize (35)-(37) as requests to stop an ongoing activity is important, because it explains (i) why the participial phrases are factual, and (ii) why'participle + per' is used here at all. As for the first point, when the activity in
question is actually taking place, then the circumstances under which it takes place, expressed by the participle, are factual, too. Compare the reverse situaation in ex. (3) in 4.1.2 above: here the modally embedded participle is non-factual, just because the action requested has not yet taken place.
From factuality we may naturally proceed to the second point: when aspeaker requests an addressee to stop doing a certain activity, the reality andimportance of this activity is likely to be foremost in the speaker's mind: 'Youare V-ing. Please, stop it. The 'you are V-ing' is so important to the speakerthat he may attach a concessive participial phrase to it ('You are V-ing, althoughyou are...') as if it is a statement in itself. But there is no statement, and the result
is that in the actual wording of the request the participial phrase acquires amodally embedded status. The concession works 'inside' the request, so tospeak.
It is clear that instances of 'participle + per' that are used as such an em bedded concessive differ in an important respect from the prototypical scalarinstances, for it is in the very nature of the scalar participial phrase to be entirely detached and modally autonomous. When used in commands, the
prototypical instance of 'participle + per' is concessive with respect to the command as a whole, in the same way as ex. (15) above.61 If the participles in (35)-
(37) are anything with respect to the command as a whole, it is causal, and thisis the best expression I can think of for the fact that they are clearly less than prototypical instances of their kind.
Entirely comparable to (35)-(37) is the following example, although itdoes not involve a negative present command:
(38) (Telemachus says that the beggar has to look after himself:)erne d'ou pôs estin hapantasanthrôpous anekhesthai, ekhonta per algea thumôi"I myself cannot possibly cope with all and sundry: I have too many
troubles on my mind." (0d. 17,13).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 147/318
PARTICIPLES I: LINGUISTICS 137
Again, the participial phrase is not concessive with respect to the statement asa whole. It may be understood 'psychologically': 'As for the beggar, he has tolook after himself. I wish I could help such people, in spite of my sorrows. Butit is just not possible/ When that to which the participle is attached is embeddedunder 'it is not possible' (ou pôs estin 62), as in (38), it is obvious that the participle, too, gets embedded in the process.
The approach in terms of modal embeddedness to 'participle + per' in(35)-(38) presented here is, I think, the best that one can make in linguistic
terms of the participles as they stand. However, I do not believe that it can goall the way to a complete explanation. For a full account of the participles, wehave to realize that they may also be described as formulas in the epic diction.Beside, and complementary to, modal embeddedness, there is a second factorin the less than prototypical status of the participles. This is the concept thatwas introduced in 1.5 above as imperfect semantic integration of formulas. Theconsequences of the participles in (35)-(38) being viewed as imperfectly integrated formulas, rather than as less than prototypical instances of the scalar participial phrase will be pursued in the following chapter (5.3).
4.4 From linguistics to the formula
4.4.1 'Participle + per' and generic modality
It is a peculiarity of Ancient Greek that active verbs may have a totally unspecified (generic) subject or object which is left unexpressed. An example isthe following relative clause:
(39) arksamenoi tou khôrou, hothen te per oinokhoeuei
"beginning from the corner where the wine is poured" (lit: "wherehe [unspec] pours the wine") (Od. 21,142).
The effect of the subjectlessness of the predicate in (39) is that all temporalreference disappears: the corner in question is identified as the place wherethe one in charge pours the wine, irrespectively of any particular circumstances.63 To this kind of generic predicate, a circumstantial participle may be attached, just as in the case of predicates with fully specified and/or referentialarguments. Consider:
(40) α ll hote tosson apên, hosson te gegône boêsas.
"But when he had come within call of the shore" (lit: "When he had
come so near to the shore as he [unspec: a 'shouter'] can makehimself audible by crying.") (Od. 5,400).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 148/318
138 LINGUISTICS AND FORMULAS IN HOMER
At first sight, the participle boêsas could be analyzed as the non-referential subject of gegône. But in view of examples like (39), which are not unfre-quent in Homer and in later Greek, boêsas in (40) had better be described asa circumstantial participle.
Now, 'participle + per' may be used in the same way as boêsas in (40), thatis, as a circumstantial participle to a predicate with (an) unexpressed argument(s) in a non-factual (generic) context. There are four examples of this use:
(41) (When Achilles has declared that he renounces his feud, theassembled Achaeans are shouting for joy. Agamemnon, beginninghis speech, cannot make himself audible:)ô philoi, hêrôes Danavi, therapontes Arêos.hestaotos men kalon akouein, oude eoikehubballein: khalepon gar epistamenôi per eonti"Friends, Danaan men-at-arms, servants of Ares. When a manstands up to speak, it is only courteous to give him a hearing and notinterrupt. The best of orators could hardly cope with such disorder."
(lit: "it is difficult [for someone], however trained he is as anorator"). (IL 19, 80).
(42) (From the same passage:)andrôn d'en pollôi homadôi pôs ken tis akousaiê eipoi? blabetai de ligus per eôn agorêtês"How can one talk or hear when everyone is making enough din todrown the loudest voice?" (Ht: "[he] is frustrated, even though he isa clear speaker"). ( . 19, 82).
(43) (Penelope to her servant, who has just told her that her husbandOdysseus has returned and is in the house:)maia phïlê, margên se theoi thesan, hoi te dunantaiaphronapoiêsai kai epiphrona per mal' eonta"My dear nurse, the gods have made you daft. It's easy for them torob the wisest of their wits." (Ht: "they are able to make [someone]senseless, however clever he is"). (Od, 23,12).
(44) (The place where Odysseus meets Nausicaa:)enth' êtoi plunoi êsan epêetanoi polu d'hudôr
kalon hupekprorheei mala per rhupoônta kathêrai."Where there was enough clear water always bubbling up andswirling by to clean the dirtiest clothes." (Od. 6, 87).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 149/318
PARTICIPLES I:
LINGUISTICS 139From the point of view of the scalarity of 'participle + per' these examples areclear and prototypical. They all involve superlativeness: the three analytic participial phrases in (41)-(43) contain an adjective denoting a property allowingof degree and comparison, and the synthetic participle of (44) is clearly of thetime-stable, adjectival kind (see 4.3.2.3 above).
However, the generic modality makes an essential difference between(41)-(44) and the other examples discussed in this chapter. This difference can
be best formulated as follows: in the examples cited earlier in this chapter, 'par
ticiple + per' is an expression in its own right, in the sense that no other ex pression with the same or approximately the same meaning could have beenused. In the case of (41)-(44), on the other hand, this is not so. On account ofthe generic modality in these examples, there is an alternative expression whichcould have been used, viz. the scalar superlative, which was discussed in 2.1.1and 3.1.1. It is easy to see that the semantic difference between the circumstantial participles in (41)-(44) and expressions involving a scalar superlative is minimal. Compare the following pair:
(45) a.blabetai ligus per eôn agorêtês. ' One is frustrated, however clear
('shrill') one can speak').b.blabetai ligus per agorêtês. 'The clearest speaker is frustrated'.
Notice, furthermore, that the participial phrase in (43) occurs in a digressivehos t e-clause, which was shown in 3.1.1 to be the favorite environment of scalarsuperlatives. And in (44), finally, the distinction between the scalar-concessive
participle and the scalar superlative is even entirely gratuite, simply becausemala per rhupoônta may be analyzed as both.
Now, it might be interesting to investigate whether there are semantic ordiscourse-pragmatic reasons for the choice of either expression. This would in
volve the determination of the conditions under which, given a generic modalenvironment, the one expression is preferred to the other. This investigationcould be interesting, but the point is that it is unnecessary in the present case,since the reason why the participial expressions in (41)-(44) have been
preferred to the scalar superlatives should not be sought in the semantic or pragmatic sphere.
It is the fact that the participial phrases in (41)-(44) are formulas conforming to a consistent localization-pattern that determines the choice. The phraseligus per eôn agorêtês in (42), for instance, is a formula which recurs elsewhere,
always in the same place in the line. The phrase fills the second part of the linein such a useful way for the poet that there is no question of there being any vi-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 150/318
140 LINGUISTICS AND FORMULAS IN HOMER
able alternative. By viewing the participial phrases as formulas, rather than assemantic or pragmatic alternatives for the scalar superlative, we have made thestep from linguistics to the other explanatory domain in the study of 'participle+ per 9 , Homeric formulas and metrics. This field will be the object of study inthe following chapter.
4.4.2 Kai, mala and strengthening
Hitherto I have ignored the occurrence of other particles than per in the
participial phrase. These particles are kai and mala. These particles appear frequently in the participial phrase, but by no means always. When we study thedistribution of kai, it quickly appears that this distribution is totally insensitiveto the distinctions I have made in the material on the basis of the prototypical
properties of the scalar-concessive participial phrase. It is easily observable thatkai just may or may not occur, irrespectively of the participial phrase being su perlative or not, modally autonomous or modally embedded.
The appropriate approach to the optional and, strictly speaking, redundant use of kai and mala in the participial phrase is strengthening. Being an ele
ment which strengthens the concession of 'participle + per', kai may becompared to even in even though.69 In being non-essential and from a strictlylogical point of view 'superfluous', strengtheners like kai may be described, interms of Prototype Theory (see 1.4), as peripheral instances of their kind. Kaias a strengthener in kai akhnumenos per does not have an independent function, as it has elsewhere.
As I will argue in 7.3.3 above, when a particle like kai is used as a'redundant' strengthener, it is less easily predictable than the more prototypical instances of its kind. Given the fact that strengtheners just may or may notoccur in a given expression-type, their distribution is more difficult to explain
than the distribution of independently expressed instances.70 This lesser predictability is in line with the fact that kai is insensitive to the distinctions discussed in 4.3.
Now one might want, in spite of the difficulties just mentioned, to justifyincidental instances of kai, by showing that the participial phrase in which it occurs has a 'stronger' concessive meaning than an instance of 'participle + per'without kai. This operation would yield, however, no positive result whatever,for in most cases by far the poet had no choice. Contrary to what may be ex
pected on the basis of the lesser predictabiHty of kai as a strengthening
phenomenon, the distribution of kai (as well as that of mala) appears to beheavily constrained by metrical factors: in most cases the presence (or absence,
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 151/318
PARTICIPLES I: LINGUISTICS 141for that matter) of kai is mandatory on metrical grounds. The rules and tendencies of the dactylic hexameter simply take precedence over the considerations
by which one might add or omit kai as a strengthener In the following chapter(5.2.2) I shall show in detail how the distribution of kai in the participial phraseis indissolubly bound up with the localization-pattern of 'articiple + per' as aword- (formula-)type in the metrical space of the dactylic hexameter.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 152/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 153/318
NOTES TO CHAPTER 4
1) It should be noted that the term 'attributive' may be used to refer to two different setsof phenomena. In the present paragraph I use the term to refer to participles that either modifya noun (like an adjective) or are substantivized. In 2.2.1 above, on the other hand, I used theterm 'attributive' in a different way, to refer to nominal predicates ('Peter's best friend' beingthe (one-place) predicate in 'Bill is Peter's best friend'). 'Attributive' in the first sense may beopposed to 'predicative', while in the second use it may be opposed to 'referential' (cf. Donellan's(1966) distinction between the referential and the attributive use of definite descriptions, seealso 2.2.2).
2) In Kühner-Gerth (1904) the term 'Adjektivsatz' is used.3) For 'complementary participles' see Goodwin (1889:347 ff.), Kühner-Gerth (1904:47-67), Rijksbaron (1984:114-119). In Greek, as well as in other languages, the verbs which take participles as a complement (i.e. as an argument, an obligatory constituent) belong to what may be called (see Kiparsky & Kiparsky 1968; Givón 1973b, 1984:117-9) 'factive cognition verbs'(know, be aware of, realize, regret, notice etc.) as well as to the 'implicative modality verbs'(manage, continue, begin, succeed, remember etc.). These verbs have in common that they either
presuppose or imply the truth of the sentence, infinitive or participial construction embeddedunder them as a complement. Notice that Ancient Greek possesses in the choice between participle or infinitive as the complement of the above mentioned verbs a means to overtly expresseither factivity (partic.) or non-factivity (inf.) by one and the same verbal lexeme. Other lan
guages have to resort to different lexemes in those cases.
For example phainomai +
participle:'appear'; phainomai + inf: 'seem'. See Goodwin (1889: 357-9), Kühner-Gerth (1904: 68-76),Rijksbaron (1984:119).
4) See Goodwin (1889:333), Kühner-Gerth (1904:77).
5) Exceptions are the subordinators hôs and epei, which may have either a causal or a tem poral understanding.
6) As regards the relation Time the situation is somewhat more complicated. The Greek participle displays an aspectual fluctuation (present vs. aorist stem) which corresponds to two possible temporal relations between the participle and the main clause: anteriority and simultaneity. See also 4.3.2.3 below.
7) There are some exceptions to this general statement. Greek participles can be mor-
phosyntactically marked as 'future' with regard to their main clause ('relative future'), in whichcase the nature of the circumstance expressed tends to be intentional or volitional, according toone of the normal uses to which the future tense paradigm can be put (see Fleischman (1982:88)). Secondly, participles may be accompanied by the modal particle an. These participles ex plicitly have either a counterfactual or potential reading, just as finite verb forms (indicative oroptative, respectively) accompanied by an. See Goodwin (1889: 70 ff.). The third exception isthat participles may be accompanied by a 'participial adverb' which marks them for one of theadverbial relations. See further below.
8) For a discussion of the discourse function of participles in Greek see Fox (1983); for participles in written English see Thompson (1983). These studies are useful, but much moreresearch is needed.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 154/318
144 LINGUISTICS AND FORMULAS IN HOMER9) Another example (beside the ones mentioned in note 5 above) of the classification is
Chantraine (1948: ii, 319 ff.). For critique on this kind of reasoning, in which translation problems
interfere with the description of the translated items (viz. the participles), see Sicking (1971: 95-
6).
10) Consequently, per in participial phrases tends to be discussed not for its own sake, but
within the context of a discussion of (concessive) participles, e.g. Chantraine (1953: 320). See
also 1.3.2 above.
11) E.g. Oguse (1962: 2,211).
12) Beside the two kinds under study here, Oguse (1962) recognizes a third kind of cir
cumstantial participle, which he calls "solidarité modale". This kind of participle occurs in imperative sentences (see 3.5.1). Here the participle should not be compared to an adverbial
subclause, but to an imperative which is co-ordinated with the imperative predicate of the
predication to which the participle is attached. Oguse's example is: apelthontes... haireisthe "Go
away and choose [lit: after you have left, choose]" (Xen. An. 3, 1, 46). I would say that the
'solidarity of the participle in such cases has to be stated in terms of sentence-type (see 3.4.1 and
3.5.1), rather than in terms of mood.
13) See Lyons (1977:797), Palmer (1986:26,51). Realis assertions are treated as unmodal-
ized when we conceive of modality as an epistemic system. But it does not follow that the dec
larative sentence-type (see Lyons 1977:745) is non-modal anyway. See 2.1.4 and 3.4.1.
14) To cite a from a discussion of modality within a logical framework (Rescher 1968): "A
proposition is presented by a complete, self-contained statement which, taken as a whole, willbe true or false. (...). When such a proposition is itself made subject to some further qualifica
tion of such a kind that the entire resulting complex is itself once again a proposition, then this
qualification is said to represent a modality to which the original proposition is subjected."
15) Professor Ruijgh points out to me that the 'autonomy* of the participle in (4) is fur
ther enhanced by the fact that it may, in addition, be interpreted as expressing absolute tense
(i.e. tense not relative to the tense-marking of the mam clause, but to the moment of speak
ing/writing): 'because they think those means of making money disgraceful'. A participle which
is meant to express absolute tense obviously possesses the maximum amount of autonomy and
detachment with regard to its main clause.
16) Both semantic and discourse-pragmatic factors, beside syntactic ones, are likely to be
important in the dependency of adverbial constituents.17) The semantic and syntactic contrast between the 'nucleus' of a sentence (the predi
cate with its obligatory arguments), where complementation obtains, on the one hand, and the
'periphery' of a sentence on the other, appears in many linguistic theories and is dealt with in
many different terminologies. For a survey of the various possibihties as well as of the criteria
used for distinguishing 'peripheral' constituents from 'nuclear' ones, see Matthews (1981: ch. 6).
18) I use even so in this study as the standard rendering of kai hôs. Accordingly, I do not
want to make any assertion about the actual use of even so as an adverbial element in English.
19) Or: may/can/will obtain. In other words, the predication may be subjected to any
'epistemic' modality (see 4.1.2 above); for the adverbial element there is no difference.
20) Adverbs or adverbial expressions can be differentiated on the basis of various criteria.
One of these is scope. Wide scope (sentential) adverbials may be opposed to narrow scope ad-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 155/318
NOTES TO CHAPTER 4 145
verbials, which modify only the predicate ('manner adverbials'). For the various types of adver bials see Givón (1984: 77-9), Pinkster (1972: ch.4).
21) Just as the Homeric demonstrative ho (later serving as the definite article) has beenreplaced by houtos. Notice that the gradual replacement of hos by houtôs leaves kai hôs untouched. In other words, in later Greek, kai hôs is a fixed expression. This is reflected in itssemantics; see 4.2.2 below.
22) In (7) hôs could in principle refer to the previous line, in which the argumentation of Neleus for hiding the horses is stated (ou gar pô ti m' ephê idmen polemêïa erga - "For he saidthat I did not yet know the practice of war"). Kai hôs then has the concessive sense which is nor
mal in later times (see 4.2.2): 'And yet... (in spite of what my father said,..)'. However, I find itmore probable that hôs refers to the circumstance that Nestor's horses were hidden from him.
23) Of course, the least likely circumstances turn into the most favorable circumstanceswhen the nuclear predication is negated, as in the case of (8), (9), (10) and (11) above. From the point of view of scalarity there is no difference. See 2.4.1 above.
24) Or one treats per as an element which emphasizes or strengthens the concessive relation expressed already by the
participle (see 1.3.2 and 4.1.1).25) See Kühner-Gerth (1904: 85), Chantraine (1948:321).
26) The fixed character of kai hôs appears also from the fact that hôs ha not been replaced by the more recent houtôs, see note 21 above.
27) See Matthews (1981: 236-39), Lehmann (1984:147) and Bakker (in prep: ch. 4).
28) Lat: qualis - talis, quantus - tantus; Gr: hoios - toios (toioutos), hosos - tosos (tosoutos)etc.
29) See Bakker (1986; in prep: ch. 3).
30) The pragmatic function 'Tail' in Functional Grammar, which is described in terms of'afterthought' (see Dik 1978:153) involves apposition.
31) As a concept, 'apposit ion' is rather elusive; it is the most undifferentiated syntacticrelation possible between consti tuents. So the discussion of apposition in Matthews (1981:224-36) is mainly concerned with the demarcation of apposition with regard to other syntactic relations.
32) See Matthews (1981: 229). The non-restrictive relative clause is often called'appositive', see Lehmann (1984: 261).
33) Hôs points ahead only very occasionally, for example Il 17,420: hôs de tis...audêsas-kon "One of the Trojans spoke as follows." But this is an odd and isolated example. See Leaf(1900-2) ad loc.
34) In Indo-European diachronic syntax, correlations in which the demonstrative elementis anaphoric (as in Latin qualis...talis) are considered to be primary ('older') with regard to thereversed forms, in which the demonstrative is cataphoric (talis...qualis). The latter are describedin Haudry (1973) as a modification of the former. The basic anaphoric nature of hôs fits in wellwith this development. In later (Attic) Greek, on the other hand, the descendant of hôs, houtôscan be cataphoric, and hence, yield correlations of the latter type.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 156/318
146 LINGUISTICS AND FORMULAS IN HOMER35) In Kühner-Gerth (1904: 85) hornôs in connection with 'participle + per' is treated on
a par with kai hôs. This is correct from a synchronic (5th and 4th century Attic), but incorrectfrom a diachronic point of view.
36) Notice that there is a third concessive adverbial element which may occur in mainclauses to which 'participle + per' is attached. This is empês. In the present study I do not discuss empês for its own sake, as the meaning of this particle does not have any bearing upon thedescription of per. However, empês does play a role in the form of formulas in which per · occurs.As such it will be discussed in 5.2.2 below.
37) There are 117 instances in the Iliad (more than 38% of the total number of instances
of per here,'post-Homeric' per (see 1.1) included), and 76 in the Odyssey (35%). See the Ap pendix.38) Concessive expressions like however much; hoewel,hoe zeer ook (Dutch) and quam
quam (Lat.) consist of the combination of superlativeness with indeterminacy. See further 6.2and 6.3.3 below. Another source of (factual) concessive subordinators are concessive conditional and temporal subclauses (viz. subclauses functioning as the sentential focus constituent ofeven): even though', quand même (French); et-si (Latin) etc. See König (1986:240). For the relation between concessive conditionals and 'concessives' see 6.1 below.
39) This means that a diachronic component is involved in this continuum, because indiachronic space scalarity precedes concession. But it does make sense to discuss the scalar participial phrase as a prototype, because the diachronic change does not yield a new expression-
type, a scalar expression-type metamorphosing into a concessive one. Rather, the change isquantitative: the scalar expression loses (some of) its characteristic properties, thereby becoming concessive. A scalar-concessive participle as in (7)-(11) maybe analyzed as a concessive participle plus some extra features. This is in line with Prototype Theory, where it is held that themore prototypical an instance is, the more prototypical properties it possesses (see 1.4). Note,furthermore, that in the case of Homer, the mapping of an essentially diachronic distinction ontosynchronic space is fully warranted anyway, because the co-existence of diachronically disparateelements is a frequent phenomenon here (see 1.5). Two items between which a diachronic relation has to be assumed can perfectly exist side by side in Homer, and it is often impossible todetermine whether they co-existed synchronically in Homer's time or not. For instance, it is a
priori impossible to maintain that exx. (7)-(11), in which kai hôs occurs in its original scalar function, cannot have coexisted in the ordinary language of Homer's time with purely concessive ex
pressions.40) The negation in oude has to be analyzed as the raised negation of the nuclear predica
tion, see note 17 of ch. 3. Occasionally, oude combines with a second negation in the nuclear predication. For examples of this phenomenon see Oguse (1968). In this case the two negationsdo not neutralize each other, as one might expect from a logical point of view. We may say thatan overt negation in the nuclear predication idiomatically triggers oude as the scalar expression,while from a strictly logical point of view kai would be possible. There are a number of exceptions to this rule (kai being used instead of oude in an overtly negated scope sentence), for example: hippos eugenês, kan êigerôn, en toisi deinois thumon ouk apôlesen. " A well-bred horse,even if it is old, does not lose its head in the hour of danger." (Soph. O.C. 25). But here the negation amounts to the expression of a positive quality ('not losing its head' being equivalent to 'stay
ing brave').
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 157/318
NOTES TO
CHAPTER 4 14741) But notice that in (16) and (19) the participial phrase stands, in addition, outside the
complement of a modality verb (see Givón 1984:120):phêmi ('I say' ('claim', 'think') and eolpa
('I assume', 'expect')). In (16) Poseidon does not mean 'I do not think that it is in spite of his fury
that Hector will not make his stand'. He means: 'I don't think that Hector will make his stand.
And I think so in spite of his fury' In other words, the participial phrase does not belong to what
Poseidon claims. The same analysis applies to (19).
42) For negative commands (of this type, in which one is requested to stop what one is
currently doing) see further 4.3.4 below.
43) In (17), for example, the fact referred to by the nuclear predication (Aeneas'
withdrawal) is in itself not very remarkable; it is even plausible, in view of the fact that his opponent receives considerable support. Likewise, in (16), Hector's envisaged withdrawal is not
remarkable in the eyes of Poseidon: he himself is helping the Greeks.
44) See Givón (1979:14, 320 ff.). See also notes 21 and 30 of chapter 2. Notice that the
three categories are gradient themselves and that there is overlap possible (see Givón 1985:211):
the most time-unstable adjective in a language may be more time-unstable than the least time-
unstable verb.
45) The distance in the continuum between substantives of the type lithos and adjectives
like khruseios is minimal, if there is any distance at all (see note 44). Notice that the latter are
often derived from the substantive denoting the material. In English (and Dutch), the adjective
cannot occur predicatively; the substantive has then to be used instead: 'This table is made of
wood' vs. 'a wooden table'.46) See also the discussion in Bakker (in prep.) of the absence of negation in Attic rela
tive and conditional clauses modified hy per.
47) Other substantival predicates in the participial phrase: Il.1, 546; 10, 548; 16, 617; 21,
483; 24,35; 24,423; 24,570; 24,609; Od. 18,21.
48) In the former case, the predicate of the ascriptive sentence is an adjective and in the
latter case a substantive.
49) In the case of meneainôn/memaôta there is an aspectual fluctuation: present vs. per
fect. This fluctuation, however, is devoid of semantic or pragmatic interest. See further 5.2.2
below (sub I.B1)).
50) Of course, sentences of the type 'Peter is drinking' are of a different type. These sentences are by no means aspectiess; they are characterized for the typically English progressive
aspect. 'Is' is not a copula and 'drinking' is non-adjectival.
51) See Kühner-Gerth (1904), Goodwin (1889: 47-9), Rijksbaron (1984: 114), Ruijgh
(1985:13-14). However, the aspectual fluctuation of (circumstantial) participles deserves an in
vestigation in its own right, in which notions like 'discourse (function)', 'backgrounding' and
'foregrounding' are likely to be important. (See for these notions note 8 of chapter 3.)
52) Actually, matters are somewhat more complicated here. The absence of degree and
measurability on the time-unstable part of the scale is itself a gradient affair (see note 44 above).
The prototypical time-unstable predicate (e.g. 'arrive') is wholly devoid of degree and superla-
tiveness, but ongoing activities can sometimes be compared to each other ('Peter worked har
der than John'). The point is, however, that 'having a certain property to a higher degree thansomeone else' is a property itself, whereas 'doing something to a higher degree than someone
else' is not an activity/event.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 158/318
148 LINGUISTICS AND FORMULAS IN HOMER53) It would be if the intended meaning of (32) is that Nestor was drunk (adjectival predi
cate). But this is not the case (see Willcock 1984: 226).
54) For 'setting the scene' as the discourse-pragmatic function of preposed subordinate
clauses, see further 6.3.1.1 above.
55) Ex. (34) is cited by Oguse (1962:135) as an instance of "association étroite" (see 4.1.2
above). But we recall that Oguse's class of 'closely associated' participles is unspecified for syn
tactic dependency and modal embeddedness.
56) Satis insolite simpliciter de causa: quoniam ("Rather unusually (per is) simply about
the cause: 'since"').
57) It is interesting to note that the causal approach to (35)-(37), or something similar toit, may be traced to remote Antiquity. In a scholion on the Hellenistic poet Apollonius Rhodius
(on i, 299, see Wendel 1935:34) it is stated that the Alexandrian scholar Aristarchus interpreted
per in the participial phrases in (77.1,131 ( = ex. (35)) and (Od. 17,47 ( = ex. (37)) as similar to
the 'evidential' particle de.
58) See also Sicking (1971: 97).
59) See also Chantraine (1948:320).
60) It is here not the appropriate place to go into the intricacies of the aspectual opposi
tion in the Greek imperative mood. Suffice it to say that present imperatives are used when the
requested action is not 'new' in the context of utterance, whereas aorist imperatives introduce
new information in the discourse. Thus, aorist imperatives are typically used after introductions
of the type 'What you must do is the following:...'. For more on Greek aspect and imperatives
(in other terminology than I would have it), see (W.F.) Bakker (1966: 31 ff.) and Ruijgh (1985:
29-38).
61) Entirely comparable to (15) is Il. 18,126: mê de m'eruke makhês, phileousa per, oude
me peiseis. ("Do not detain ( = stop detaining me) from the battle, however much you love me.
For you will not persuade me.") Here the participle is not concessive with respect to the ongo
ing activity, but with the request as a whole. With respect to the ongoing activity, the participle
may be compared to a causal subclause: you are detaining me because you love me. Thus, as
regards the respective functions of cause and concession, this example is a complete reversal of
(35)-(37).
62) In later Greek this is expressed as ouk estin hopôs (lit: 'there is no way how').
63) Notice that subjectless predicates may also be used when the agent of an activity either
is unknown or totally unimportant in the context. An example of the latter alternative is the tem
poral subclause epei esalpinxe ("when the trompet sounded" - lit. "when he [unspec] sounded
the trompet") Xen . An. 1,2,17. Here the activity is specific: someone sounded the trompet at a
particular moment, but the identity of this person is unimportant.
64) Moreover, as professor Ruijgh points out to me, substantivized participles occur in
Homer nearly always with the definite article: ta eonta ('the (now) existing things').
65) In (41) the participial phrase contains, instead of an adjective, a synthetic participle
of the adjectival kind (epistamenos).
66) In Il. 2,246 and Od. 20,274. The phrase conforms to a pattern itself. Ligus and agorêtês
may be substituted for: thoo sper eôn polemistês ('quick warrior as he is'), see ex, (17) above.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 159/318
NOTES TO CHAPTER 4 149
67) Notice, however, that after Homer kai is always present in the participial phrase, forming with per the petrified collocation kaiper.
68) Notice that the situation with mala is different: conforming to its meaning ('very'), this particle occurs only with superlative participles (mala per memaôtalrhupoônta etc.).
69) Or ook in ook al in Dutch.
70) By this I do not mean that the distribution of stengtheners cannot be explained anyway.In fact, subtle discourse-pragmatic factors may be at work in the distribution of strengtheners.But these factors are more 'fuzzy' and less straightforward than the factors determining the distribution of prototypical instances. In addition, the characteristics of the personal 'style' of a
speaker or writer are likely to be more apparent in the peripheral instances of a given item thanin the more 'objective' prototypical instances.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 160/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 161/318
5 PARTICIPLES : FORMULAS AND METRICS
5.0 Introduction
After the linguistic analysis of 'participle + per' in the previous chapter,we are now concerned with the remaining features of this expression-type. Inthe present chapter I will discuss 'participle + per' in connection with two majorcharacteristics of the Homeric text, its metrical form and its oral-formulaic nature. These two features determine to a high degree the linguistic form of the
Iliad and Odyssey and they will appear to have important implications for thestudy of 'participle + per', too.
The chapter consists of three parts. The introductory part (5.1) is a discussion of the 'Homeric formula'. The point of view adopted for this discussionwill be primarily linguistic. I will treat formulas as normal ordinary languageexpressions which have a specific function in the oral-formulaic diction. I willargue that an account of the Homeric formula in terms of function has an advantage over the accounts in which the formula is treated as a phenomenon sui
generis, for which a special explanation and definition has to be sought.
In 5.2 I discuss the instances of 'participle + per 9 as to their (metrical)
form. I will show that the form of 'participle + per 9 may be adapted by the poetto the exigencies of the hexameter line: the metrical word-type to which a giveninstance of 'participle + per' belongs may be altered in systematic ways by theaddition of certain particles and other elements whose meaning is subservientto their metrical-extending function. I will show that the rationale behind thisaddition lies in the usefulness of 'participle + per' for easy and smooth versification. The expression-type forms a formular system which is essential forthe rapid and automatic processing of rhythmically and linguistically acceptable hexameters.
In section 5.3I deal with the meaning and function of 'participle + per 9 in
the formulaic diction of Greek epic. In this section, I will introduce the notion
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 162/318
152 LINGUISTICS AND FORMULAS IN HOMERof 'integration of formulas'. It is a typical consequence of formulaic languageuse that phrases may be less integrated' in their context, precisely because theyare used as formulas, viz. as ready-made phrases which facilitate the poet'sautomatic filling the metrical space of the hexameter line with words and
phrases. Formular (non-)integration may occur at various levels', ranging from phonetics to the content of whole passages. The level which is relevant for thestudy of 'participle + per' in this connection is semantics.
5.1 On formulasSince the Parryan revolution in Homeric studies, most discussions of
Homeric language and style may be grouped around two central issues: (i) howis the concept of 'formula' to be defined, and (ii) does formularity entail orality,or, in other words, is the presence of formulas in the Homeric text to beregarded as an index of the oral composition of that text? Milman Parry wasthe first to raise these issues, which have been vexed questions ever since his
publications. In his first French thesis (1928a), Parry proved that the well-known noun-epithet expressions in Homer form a formular system which is
traditional, rather than the invention of a single poet. Later (1930, 1932), hemade an important step further, by contending that Greek epic poetry is notmerely 'traditional', but also that Homer as we have it is orally composed.Homer's orality, he argued, can be deduced from his formularity.
Now, it is important to realize that Parry's conception of Homer's oralityis intimately connected with what we consider to be a formula and what not. Inother words, the second question mentioned above heavily depends on the first:if we take it as a premiss that formularity entails orality, then the broader thedefinition of 'formula' we accept, the higher the degree of formularity inHomer, and the stronger the case for Homer as an orally composed text. Ac
cordingly, when Parry made the step from 'traditional' to 'oral' poetry, he hadto considerably extend his definition of 'formula'. Out of the original definitiongrew a rather fuzzy notion which was more and more extended by Parry's followers, all for the sake of proving Homer's orality (see further below, 5.1.3).2
Parry's well-known definition of 'formula' runs as follows (1971: 272 =1930: 80): "a group of words which is regularly used under the same metricalconditions, to express a given essential idea."3 There are both 'extensional' and'intensional' problems with this definition. Extensional problems arise whenwe try to delimit the formular from the non-formular. There are a great many
acknowledged, prototypical formulas, but there are many peripheral cases aswell, to which the above definition does not or not fully apply. In any case, Parry
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 163/318
PARTICIPLES Π : FORMULAS AND METRICS 153
and his followers have called some phrases a formula which satisfy none or fewof the conditions of the definition.4
The intensional problems arise when we try to define what it means for agiven phrase to be a formula. The question which arises here is not how usefulParry's definition is when applied to the doubtful and peripheral instances offormula', but how appropriate it is as a characterization of the prototypical
cases. The lack of agreement about the definition and the continuing confusionabout the concept of formula in general has been one of the most salient features of oral poetry studies. As Austin (1975: 14) has put it, "The scientists,from whose vocabulary we have borrowed the concept of formula, might beastonished to find Homeric scholars exercising ideosyncracy at the level of
primary axiom and yet reaching unanimity at the level of general theory." Nobody denies that Homer is full of formulas, but on the question as to howfull there is no agreement.
The trouble with the approach to orality instigated by Parry is that it induces a 'structural', context-independent account of the formula: when the
presence of formulas in a text is considered to be an index of the oral composition of that text,5 then it is of the highest importance to know how to recognizeformulas as such, in other words, to know what a formula is. Very often, theclassification (whether or not controversially) of a certain phrase as a formulahas implied that the phrase was considered a formula in itself, an entity that can,
by some inherent property, be differentiated from non-formular phrases. Themethod by which one starts with (a definition of) the formula and proceedsfrom there to conclusions about oral composition, typically reflects the cognitive development of the researcher, who observes some phenomena (i.e. for
mulas in Homer), tries to conceptualize them and subsequently explain themand/or draw conclusions about them.
But the point is that formulas are not a phenomenon sui generis. Formulasare not there just for their own sake; their presence is motivated by the exigencies of oral composition and performance. Accordingly, afunctional approachto the Homeric formula seems to be more promising than a structural approach.It is not the case that certain phrases are formulas; rather, they are used as formulas. This approach differs from the approach of Parry in that we start withorality and proceed from there to the formula. In this approach, we take theoral composition of a text as a hypothesis and see whether it has left any traces
in the text. Such traces may be formulas of the prototypical kind, but also phenomena that need not, or cannot, be discussed in terms of 'formula'. The
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 164/318
154 LINGUISTICS AND FORMULAS IN HOMER
advantage of this approach is that the heavy pressure on the definition offormula' disappears, as there is much less that depends on it.
In the following subsections, I will try to give a functional account of theHomeric formula which points ahead to the discussion of the peculiarities of'participle + per' in the oral-formulaic diction. In 5.1.1 I will discuss the meaning of formulas, on the basis of Parry's notion of an "essential idea" which isexpressed by a formula (see the definition above). In 5.1.2 the metrical dimension of formulas will be dealt with, and in 5.1.3 the recurrence aspect.
5.1.1 Formular language and ordinary language
In a primarily semantic study it is apposite to start the discussion with the part of Parry's definition that is concerned with meaning: according to Parry, aformula is used to express a given essential idea. From a semantic point of view,the term 'idea' is unsatisfactory.7 But it is more important to focus on thesemantic implications of Parry's definition.
Parry explains 'essential idea' as follows (1971: 13): "What is essential inan idea is what remains after all stylistic superfluity has been removed." Thus,
in the example which Parry provides, the noun-epithet formula polutlas diosOdysseus ('much suffering, godlike Odysseus'), the essential idea is 'Odysseus'.Likewise, the essential idea of the formular line
(1) êmos d' êrigeneia phanê rhododaktulos Eos
"As soon as Dawn with her rose-tinted hands had lit the East."
is 'when day broke'.The implication of Parry's distinction between formulas and the 'essen
tial idea' expressed by them is that formulas and language use involving formulas is qualitatively different from non-formular, normal language use.Parry's distinction amounts to the equation of the essential idea of a formulawith what the formula actually means in terms of ordinary language.8 The essential feature of the formular language as Parry defines it, in which formulasexpress 'essential ideas' is that form (expressions, formulas) is separated frommeaning (essential ideas) to a much higher degree than in non-formular language.
Parry's account of 'formula' must be considered against the backgroundof the way the Homeric language and diction was viewed in the first decadesof the century. In Witte (1913) and Meister (1921), the language of the Homeric
poems is considered to be, and systematically described as, a language in itsown right which is not created by any one poet, but by the medium in which he
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 165/318
PARTICIPLES Π : FORMULAS AND METRICS 155
composes, viz. the epic verse. This hexametric language was considered a reallanguage, a Kunstsprache which is distinct from ordinary Greek. TheKunstsprache is characterized by a large number of archaic, dialectal and evenwholly artificial elements, whose presence may be justified by the poet's constant need to adapt the expression ('form') of his thought to the complicated exigencies of the hexameter line.
Parry added a new and essential dimension to this approach, in arguing,and proving, that the deviations from ordinary language are not there merely
for the sake of metre and the Kunstsprache. Parry showed that these deviationshave their motivation in the fact that the Homeric diction is, or in any case wasoriginally, designed, over many generations of epic poets, to facilitate oral com
position. For all its artificiality, the Homeric diction is functional in the first place. Parry discovered the pragmatics of the Kunstsprache, so to speak.
One of the most conspicuous features of this Kunstsprache is the systemof noun (name)-epithet formulas. Parry proved that these formulas play an im
portant role in the oral-formulaic diction in that they enable the poet to referto a hero or a god in an entirely systematized way. The epithets are chosen andgrouped in accordance with the case form in which the noun/name is to appearand the metrical position it is to occupy. It is precisely this kind of formula towhich the rigorous distinction between 'form' (ulila) and 'essential idea' best ap plies.
The distinction between 'formula' and 'essential idea' has important consequences for the semantics of the formula as Parry defines it. Owing to the
presence of ornamental, purely formal material, the meaning of the formulacan be conceived of in dualist terms. It can be defined without reference to theform(ula) itself. It is important to realize that this is different in ordinary language. In ordinary language, form and meaning are distinct too, of course, but
they cannot be so easily separated as Parry does with his formulas in theKunstsprache. Very often, the 'idea' which a given expression expresses,whether it is essential or not, is not separable from that expression itself. Accordingly, the meaning of a given expression cannot easily be defined withoutreference to the form of that expression.
Now, it might seem that Parry's distinction between 'meaning' and 'form',which is absent in the ordinary language, at least to that high degree, constitutesthe delimitation of the formula with regard to what is non-formular. However,this is not so. The separation of form and meaning applies by no means to allthe phrases that have been called 'formulas', by Parry himself and others. Thismeans that Parry's definition of 'formula' does not apply to all the cases to which
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 166/318
156 LINGUISTICS AND FORMULAS IN HOMER
it is meant to apply. The implication, in a wider perspective, of this is that wemay ask whether the Kunstsprache to which the noun-epithet formula belongs,is really so all-pervasive a feature of Homeric poetry as we are led to believe.
Beside the noun-epithet formula polutlas dios Odysseus and cases like ex.(1) above, Parry gives another example of a formula and its essential idea (1971:14):
(2) ton d' aute proseeipe"and to him spoke."
According to Parry, the essential idea of this formula is 'said to him', just as'Odysseus' is the essential idea of polutlas dios Odysseus. But there is a difference between the two. We cannot possibly explain the formular status of ex.(2) on the basis of its being the expression of an 'essential idea', for in doingthis, we characterize in the last resort everything as a formula. The essential ideaof (2) is (2) itself. Like any other natural language expression, (2) cannot bedissociated from its 'essential idea', viz. its meaning.
Yet, in spite of its being different from noun-epithet formulas like polut-las dios Odysseus in that its meaning cannot be separated from its form, (2) isno less a formula than a noun-epithet expression. Together with a noun-epithetformula, it forms wholly stereotyped, formulaic Hnes: being a so-called T1-for-mula (a phrase filling the first half of the line, before the trochaic caesura), itcan be complemented (both in a metrical and in a syntactic sense) by phraseslike polutlas dios Odysseus, which belong to the T2-formulas, filling the slot between the trochaic caesura and the end of the line.
It follows, then, that as regards semantics we may distinguish two kinds offormula: the formula whose meaning is distinct from its form and the formula
12
whose meaning is not. Parry's definition is primarily based upon the first kind,of which the noun-epithet formulas are the prototypical examples. The noun(name)-epithet formula is characterized by what will be called henceforthredundant semantics: the element that constitutes the divergence of form andmeaning, viz. the epithet, is redundant from a semantic and pragmatic point ofview in that it is unnecessary for the identification of the hero in question.13 In5.2 I will deal with phenomena that should be studied in connection with theredundant semantics of the noun-epithet formula.
Noun-epithet expressions are the best examples of what is commonlyknown as the Homeric formula. But it is important to emphasize that their for
mular status is not due to some inherent property: the mere addition of anepithet to a name does not yield a formula. The addition must also be function-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 167/318
PARTICIPLES Π : FORMULAS AND METRICS 157
ally motivated. The functionality of noun-epithet formulas is that they conformto a consistent pattern which greatly facilitates oral composition and performance. This applies equally to the second kind of formula, which is not charac
terized by the redundant semantics of the noun-epithet formula. Thus, when itcomes to defining the concept of formula, the account in terms of an "essential idea", which is left "after all stylistic superfluity" has been removed, losesits determinative force.
Parry was the first to discover the functional motivation for the presence
of formulas in Homer. Yet he did not carry this discovery to its logical conclusion. In treating phrases as formulas because of what they are or mean,
Parry (and those working in the line of research he instigated) missed a chanceto arrive at an account of the formular use of language which is plausible from
a linguistic point of view. Instead of being qualitatively different (by some inherent property) from ordinary language phrases, formulas are what they are
because of the way in which they are used. Thus I adopt a functional frameworkfor the study of formulas; in Parry's approach, on the other hand, the notion offormula is a category, rather than a function.15
The relation between formular language (or language containing formulas) and ordinary language,16 has received much less attention than theother aspects of the formula, viz. its metrical dimension and its being a repeated
phrase (see the following subsections). The reason of this seems to be that the
formula has been mostly studied, within an exclusively oral-formulaicframework, as a phenomenon sui generis which has to be accounted for. Thequestion as to what a formula is has amounted in practice to the problem howto define the formula qua most characteristic feature of the diction. Accordingly, the question whether the oral formula has any connections with what isoutside the formulaic diction, viz. with ordinary language, has not often been
asked.17
In the present chapter I shall try to illustrate on the basis of the expression-type 'participle + per
9 what it means for an expression to function as a
'formula'. In other words, I shall discuss the consequences of the fact that 'participle + per' as a modally autonomous, syntactically independent, and henceoptional constituent, is useful for the poet as a formula. This opens the way fora different approach to the Homeric formula: to start from ordinary languageand see in what ways an ordinary language expression can be useful in the for
mulaic diction.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 168/318
158 LINGUISTICS AND FORMULAS IN HOMER5.1.2 Formula and metre
We now turn to the metrical dimension of formulas. The metrical aspectof Homeric formulas is connected with the subject of the next (5.1.3) subsection, recurrence. According to Parry's definition, a formula is "an expressionwhich is regularly used under the same metrical conditions." The 'sameness' ofthe 'metrical conditions' obviously implies recurrence. Apart from this, metreis in itself a kind of recurrence. But the subject of the present subsection is intimately connected with the previous subsection, too, since the distinction be
tween formulas whose meaning is distinct from their form and formulas whosemeaning is not, has a metrical implication.
When one works within the framework of the Kunstsprache, the metricaldimension of formulas comes as a matter of course. The presence of many artificial and dialectal forms is ultimately due to the influence of the verse, asWitte (1913) and Meister (1921) and others before them have shown. In thecase of the noun-epithet formulas, we may say that metrical form is often theidentity of the formula, in that it motivates the addition of the epithet: anepithet is added to a noun or name so as to fill a given slot in the hexameter; if
the name of a particular god or hero is to occupy another slot and/or is to ap pear in another case form, another epithet is selected accordingly. Thus, in thechoice of the epithet, the semantically redundant element, the poet is heavilyconstrained by the metrical structure of the verse.
In the case of the second kind of formula distinguished in 5.1.1, whereform and meaning are not so distinct (see ex. (2)), metre is an important factor, too. But here it is not so much a factor by which the poet is constrained inthe addition or omission of semantically redundant material. Rather, we haveto conceive of the rhythmical form of a phrase like (2) as a necessary conditionfor its being used in the metrical space of the dactylic hexameter. The phrase
in (2) can be used as a formula on account of its metrical form; it has an ap propriate length and the kind of rhythmical characteristics that make it a useful phrase for the dactylic hexameter. But it does not contain elements whoseexistence is solely due to the influence of the verse. There is no question of its
being adapted to the exigencies of the verse in any way; it merely satisfies thoseexigencies.
Necessary conditions are always negative. When we say that an ap propriate rhythmical shape is a necessary condition for the use of a given phrasein the metrical space of the dactylic hexameter, this means that that phrase
could not possibly have
been used in the hexameter (whether or not functioning as a formula) if it did not have the appropriate rhythmical shape. But rhyth-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 169/318
PARTICIPPLES Π : FORMULAS AND METRICS 159
mical appropriateness does not in itself entail formularity; in other words, it is by no means a sufficient condition for formularity. The participial phraseakhnumenos (teiromenos, hiemenos) per, for example, has a very apt rhythmi
cal form: it can be conveniently used to fill the slot between the so-called'bucolic diaeresis' (see 5.2.1.1 below) and the end of the line, functioning as aclosing cadence (clausula).19 But this fact alone does not make of akhnumenos
per a formula.
Again, we have to discard a context-independent property of a given
phrase as the reason why the phrase is formular. In the previous section, thiswas 'meaning'; in the present one it is the metrical form of a phrase and the place it occupies in the verse. Metre is not an active structuring principle, determining the actual form of phrases and formulas. Rather, we have to say that the
poet constantly adjusts his choice of words and phrases to the metrical possibilities. Even in the case of the noun-epithet formulas, where it is metre thatseems to determine which epithet has to be used, it is the poet who adapts the
phrase to the metrical possibilities. His choice of the epithet is constrained insystematic ways, and this is precisely the basis of the functionality of the noun-epithet formular system.20
5.1.3 Types of recurrence
'Repetition' is an important aspect of the Homeric style, and it has playeda major role in the controversies about how to define the formula. There ismore than one kind of repetition. Repetition may involve the exact, verbatimrecurrence of a given phrase, but it may also involve the recurrence of a rhythmical pattern. These two possibilities may be seen as the extremes of a con
tinuum, in which repetitions may be ordered by degree of linguistic sameness.
This continuum ranges from minimal linguistic sameness (mere metrical
repetition) to maximal linguistic sameness (verbatim repetition). In between,we may localize repetitions with linguistic sameness on the level of category
('parts of speech').From the discussion of the metrical aspect of formulas in the previous sub-
section, it follows that mere metrical repetition, the recurrence of rhythmical patterns, cannot, and should not be considered the specific property of formularity. If mere metrical repetition is considered to be formulaic, then every
thing, both in Homer and in later hexameter poets is a formula, simply becausein these poets everything is metrical. Metrical repetition without reference to
formulas has been the subject of a well-known study in Greek hexametermetrics, O'Neill (1942). O'Neill showed that there are severe constraints on
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 170/318
160 LINGUISTICS AND FORMULAS IN HOMER
the placing of words (regardless of their meaning and seen purely as metricalword-types), especially the longer ones, in the metrical space of the Greekhexameter. O'Neill's data reveal that often there are only one or two possibilities to localize' a given word in the verse. As O'Neill was exclusively concerned with the presentation of the data involved, it was left to others (e.g.Beekes 1972) to 'interpret' and explain his data.
Further on (5.2.1), I will discuss O'Neill's findings in metrical localizationin more detail. For the moment, I am more concerned with the kind of repeti
tion that involves linguistic sameness. When one starts the discussion of formulas with verbatim repeated phrases and allows the definition of 'formula' tocover more than those phrases, one has to introduce the notion of substitution.In this approach, one part of a phrase, filling a given metrical slot, is seen as avariable whereas the rest of the phrase is treated as a constant. This was thestep taken by Parry after his two principal French theses. The classic exampleof substitution (1971:313) is aphrase filling the metrical space after the bucolicdiaeresis (a so-called 'bucolic clausula'):
(3) alge' ethêke: - ⋃ ⋃ - ⋃
("brought sorrows")Here we may substitute for the noun (alge(a)) another noun of the same metrical form and take the verb as a constant:
(4) kudos ethêke("brought renown")
Or we may, conversely, take the verb as variable and substitute another verbfor it:
(5) alge'edôke("gave sorrows")
This substitution system can be characterized as the combination of verbatimrepetition (the constant) with repetition on the level of category (the variable).Parry limited his conception of 'formula' to this level of linguistic sameness.But he showed the way for a more general treatment, in stating (1971: 313),among other things, that teukhe kunessi, a bucoUc clausula from one of theopening lines of the Iliad, is "like" dôken hetairôi. This amounted to allowingsubstitution with two variables and repetition wholly on the level of category.
But Parry went another step further, in contending that "often one finds the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 171/318
PARTICIPLES Π : FORMULAS AND METRICS 161
same verse-pattern where the words are different" (1971: 313). Parry did not
actually call the following two lines two instances of one and the same formula,
(6) nouson ana straton orse kakên, olekonto de laoi"He inflicted a deadly plague on the army, and the men died" ( Il. 1,10).
paida d' emoi lusaitephilên, ta d' apoina dekhesthai"Release my daughter and accept this ransom" ( . 1, 20),
but he is near to it. The repetition in (6) approaches the level of minimal linguistic sameness and I find it difficult to conceive of any functional motivation for it.
Just as in the case of the two lines in (6), Parry did not go as far as to callteukhe kunessi and dôken hetairôi two instances of one and the same formula.This step was left for others to make. Once it was made, the concept of 'structural formula' came into being (Russo 1963, 1966), in which the border-line
between the formulaic and the non-formulaic in the continuum between meremetrical repetition and verbatim repetition was situated at the point where we
may speak of repetition (and hence identity) in category. Thus the phrases in(3)-(5) would be instances of the following structural formula ( N = 'noun', V= Verb'):
(7) N V 12- ⋃ | ⋃ - -,
whereas Parry's pair teukhe kunessi and dôken hetairôi would be two instancesof a structural formula in which the two parts of speech are reversed:
(8) V N 12
- ⋃ | ⋃ —Another structural formula which Russo identified was a word-pattern
consisting of a middle participle placed at the beginning of the line, filling thefirst foot and a half. According to Russo (1963:242), this is the ultimate, unifying, reality behind cases like
(9) oulomenên, he muri'Akhaiois alge' ethêke"the fatal (wrath) which brought the Achaeans many sorrows" ( Il.1,2) 24
akhnumenos: meneos de mega phrenes amphi melainai"(Agamemnon leapt up.) His heart was seething with black passion"( Il.. 1,103)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 172/318
162 LINGUISTICS AND FORMULAS IN HOMERThis conception of the formula, where the essence of 'formula' lies in the simultaneous recurrence both on the level of metrics and on the level of category,has provoked considerable reaction (see Hainsworth 1964, Minton 1965,Hoekstra 1965:14-6). There were two main objections to the concept. First, inthe conception of 'structural formula', the definition of formula, it was thought,is watered down to the point of vacuoussness: how are we to distinguish between the formulaic and the non-formulaic when we allow such a large amountof unpredictable variability?25 Second, the kind of phrase-patterns to which the
concept of structural formula applies are not so much characteristic of Homeric poetry specifically, as of Greek hexametric poetry in general. Thus, on Russo'saccount of the formula, we would be unable to differentiate between Homeric
poetry and unequivocally literate poetry, whereas such a differentiation shouldhave the very concept of formula as its basis.
This controversy is obviously relevant for the present study. Instances ofthe expression-type 'participle + pef placed at the end of the verse clearly haveall the characteristics of what Russo calls a structural formula: they involverecurrence both on the level of metrics ('bucoHc clausula') and on the level ofcategory ('participle + pef). As regards localization and 'structural
formularity', the scalar-concessive akhnumenos per as a clausula at the end ofthe line can without more ado be compared to the non-scalar akhnumenosfunctioning as an instance of Russo's choriambic (word-type - u u - ) middle
participle placed at the beginning of the line (see (9) above.Hoekstra (1965:15,25) argued that single words are not to be called for
mulas, on the grounds that if they are, the concept of formula loses all its ap plicability. If this ban upon the one-word formula is justified, then akhnumenos per is not to be called a formula. In Hoekstra's account it is simply the productof metrical localization, and as such wholly devoid of formular interest. The
same would apply to the scalar-concessive participial phrases of other metricalshapes; we would have to do with word-types, not with formulas.In the next two sections (5.2 and 5.3) I shall argue that Hoekstra's fear of
the one-word formula is, generally speaking, unnecessary. To my mind, thenumber of words, one, two or more, of which a phrase consists, is immaterialfor that phrase being or not being a formula. Formularity does not depend onsuch a formal, context-independent property. And participles of one and thesame word-type (whether or not accompanied by per) that are placed in a fixedslot of the hexameter are, in Homer, definitely more than just a case of localization.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 173/318
PARTICIPLES Π : FORMULAS AND METRICS 163
But this does not mean that Russo's concept of 'structural formula' is healthy and free of problems. It seems to me that the objections against Russo'sconcept were entirely justified in so far as Russo contended (1963: 240) thathis localized word-category-types are in themselves formulaic. In doing this,
he weakened his position considerably, for what is in itself formulaic is no lessformulaic when a Hellenistic pen poet writes it than when the supposedly oralHomer uses it.
The inherently formulaic is in the last resort formulaic independently of
the medium for which it makes sense to speak in terms of formulas at all, viz.oral-formulaic poetry. Clearly, Russo's structural formula is an extension of the
kind of approach in which formulas, or what are considered to be formulas, arestudied as to their frequency and recurrence, in an entirely non-functional,mechanical way. Parry himself had instigated this kind of approach, in his well-known formulaic analysis of the first 25 lines of the Iliad and the Odyssey (1971:
301 ff.), in which he studied the recurrence of formulas and formulaic phrases
on various levels, see (3)-(6) above.In the previous subsections I argued that it is not the case that formulas
are formulas solely because of their meaning or their metrical form. This claimmay now be repeated in connection with recurrence and Russo's structural for-
mula. A participle like akhnumenos placed at the beginning of the Une orakhnumenos per placed at the end is not a formula solely by virtue of the factthat it is a participle conforming to a consistent localization pattern. It is a for-mula because it is used as a formula. The recurrence and the localization are
28
simply consequences of that use; they do not in themselves constitute it.
In considering certain phrase-patterns as in themselves formulaic, Russostudied the phenomenon from the outside, so to speak. If a phrase is to be called
formulaic ór a formula, its repetition has to be functional. The repetition of a phrase or phrase-pattern can only be conceived of as an index of the formularityof that phrase(-pattern) when we may speak of a recurrent and systematizeduse which facilitates, and is in function of, the rapid and easy filling of the metrical space of the hexameter with words and phrases which are meant to performa narrative function.29 Russo's structural formulas should be assessed on the
basis of the functionality criterium, and denied formular status if they do notserve this end in any unequivocal and perceivable way.
In the previous subsections (5.1.1 and 5.1.2) I made a distinction betweenformulas whose meaning is characterized by 'redundant semantics' and whose
form is an adaptation on the part of the poet to the exigencies of the hexameter,on the one hand, and formulas which do not in themselves differ from ordinary
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 174/318
164 LINGUISTICS AND FORMULAS IN HOMER
language expression, on the other hand. The structural formula as discussed byRusso obviously belongs to the second kind. The analysis in terms of substitution in one or more variable slots, on which the concept of structural formularests, can be applied, too, to ordinary language. In ordinary language we mayconceive of, for example, the subject position in a sentence or the focus constituent of a scalar particle as a variable for which, within a given set of selection restrictions, alternatives may be substituted. This mechanism differs notqualitatively from substitution in a structural formula, as exemplified by exx.
(3)-(5).30
Recurrence is the common denominator of both kinds of formula. And in both cases it is a consequence and an index of the usefulness of a given phraseas a formula. In the following two sections, I discuss 'participle + per' in connection with the two kinds, respectively. In 5.2 I discuss 'participle + pef inconnection with 'redundant semantics' and the noun-epithet formula. Here wewill have an instance of an expression whose form is adapted by the poet in systematic ways to the exigencies of the hexameter. And in 5.31 discuss 'participle+ pef as an ordinary language expression. Here we will deal with what mayhappen when an ordinary language expression(-type) is useful as a formula.
5.2 Formulas and form: the localization of formulas
In this section I discuss the rhythmical form of the various instances of'participle + per', and the place which they occupy in the verse. This discussion involves a digression into metrics. Its upshot will be the explanation of thedistribution of kai and other particles in the participial phrase, as well as an insight in Homeric versification and the role of 'participle + pef therein.
Speaking in terms of types of recurrence (see 5.1.3 above), the kind ofrepetition which is involved in the case of 'participle + per' is repetition on the
level of category. The great majority of the instances of 'participle + pef can be divided into a number of 'structural formulas', in which per is the constantand the participle (or, in the case of analytic participles: the adjective) the variable.
The discussion of 'participle + pef in connection with metrics is a confirmation of the statement that if repetition is to be functional (viz. have formularinterest), it should not involve too less linguistic sameness (i.e. be too abstract).In 5.2.2 below, I shall show that the instances of 'participle + pef form whatmight be called a formular system. The functionality of this system in the dic
tion depends on the linguistic identity of the phrases that constitute the system.A given instance of 'participle + pef, which fills a rhythmically salient part of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 175/318
PARTICIPLES Π : FORMULAS AND METRICS 165
the verse in a standardized manner, may have the same rhythmical propertiesas an instance of an entirely different expression-type, but this does not meanthat the two phrases are to be called one and the same formula (see also 5.1.2).
Nor do the two phrases belong to one and the same formula-system. Meremetrical similarity is not enough for formular similarity. Formular sameness isalways concomitant with some kind of linguistic sameness.
The usefulness of the formula-system 'participle + per' for easy andsmooth versification is greatly enhanced by the linguistic properties of 'par
ticiple + per 9
. The syntactically detached status of 'participle + per' (considerthe prototypical properties of independence and modal autonomy), leads to avery flexible formula-system, because 'participle + per
9 is always optional. Anyinstance can be easily suppressed whenever the metrical space is preferably to
be filled with other words. Conversely, 'participle + per 9 can be without dif-
ficulties added to a given sentence even when this addition is unnecessary andstrictly speaking 'redundant'.
In the following section (5.3), I will discuss the semantic aspects of theredundance which is the consequence of automatic formular use (see 5.3.2). Inthe present one I am concerned with the metrical and rhythmical aspects of theformular use of 'participle + per 9. But before these aspects can be discussed,it is useful to provide their metrical background.
5.2.1 The dactylic hexameter
In O'Neill (1942: 105), a distinction is made between the 'outer metric'and the 'inner metric' of the dactylic hexameter. The outer metrical featuresof the hexameter are that it consists of six feet. The dactyHc foot consists ofthree syllables: 'long + double short'. The alternative realization of the foot isthe spondee, in which the double short element is replaced by a long element
(two syllables: 'long + long'). Thus the distribution of spondees over the verseis an outer metrical affair. The 'outer metrical' facts of the hexameter constitutewhat may be called the metric profile of the verse (see van Raalte 1986: 29).Inner metric, on the other hand, involves the words in the verse and the waysin which the metrical space is filled by the word-types of which the words areinstances. Here we are in the domain of caesuras (places where word end isobligatory or desired) and 'bridges' (places where word end is forbidden oravoided). The 'inner metric' of the hexameter is what may be called the rhyth
mical structure of the verse. In the present study we are concerned with the
rhythmical structure of the hexameter and the rules and tendencies obtainingthere.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 176/318
166 LI NGUISTICS AND FORMULAS IN HOMER
In studying a representative sample of hexameter lines of poets fromHomer down to the Alexandrians, O'Neill (1942) showed that throughout thehistory of the Greek hexameter there are severe constraints upon the localization of word-types, especially the longer ones, in the metrical space. In orderto unequivocally tabulate the quantitative data associated with his research,O'Neill devised a reference system (which has become standard), in which eachmetrical position is assigned a number. This system yields the following schemeof the metric profile of the hexameter:
(10) 1 1.5 2 3 3.5 4 5 5.5 6 7 7.5 8 9 9.5 10 11 12-⋃ ⋃ -⋃ ⋃ -⋃ ⋃ -⋃ ⋃ -⋃ ⋃ - ⋃
Using this scheme as a point of reference, we can present the localization of,for example, the word-type ⋃-⋃⋃- as follows (see O'Neill's table 20, 1942:146): it occurs ( = ends in) in the Iliad 28.3% in position 5, and 67.5% in pos.9. For the Odyssey these figures are even complementary (23.5% and 76.5%,respectively), the word-type occurring nowhere else in the sample.
O'Neills quantitative research yields figures which are in themselves valuable and important, but it suffers from two major deficiencies. First, as he himself admits (1942:106), O'Neill does notexplain his data, and second, the notionof 'word', the ultimate object of localization, is more elusive than appears fromO'Neill's treatment. Both deficiencies are of importance for the discussion ofthe localization of 'participle + per' presented below (5.2.2); the two following subsections (5.2.1.1 and 5.2.1.2) deal with them.
5.2.1.1 Rules and tendencies
O'Neill's localization figures acquire real significance only when they areseen as the consequence of the rules and tendencies that constitute the rhythmical structure of the hexameter. In Beekes (1972), these rules and tendencieshave been described. This description has yielded an interpretation of O'Neill'sdata, thereby providing what seems an adequate explanation of the localizationof word-types in the metrical space of the hexameter.
Beekes argues that all word-types are localized according to a limited setof metrical rules and tendencies. He presented these rules as follows:
(11) 1) Word end is (near-)obUgatory at either 5 or 5.5 (main caesura:'penthemimeres' (henceforth: 'P') and trochaic caesura ('P')).
2) Word end is desired at 8 (bucoHc diaeresis).3) Word end is forbidden at 7.5 (Hermann's bridge).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 177/318
PARTICIPLES II FORMULAS AND METRICS 1674) Word end is avoided at 3.5 and 4 after a long final syllable
(Meyer's First Law).5) Word end after a long final syllable in 8 is strongly avoided and
forbidden after a long final syllable in 10.6) Word end is avoided at 11.
On the basis of these rules we may now present the scheme of the rhythmicalstructure of the hexameter (whereby '|' denotes obligatory word end(caesura), '|' denotes desired word end, and'∩ ' denotes forbidden or avoidedword end (bridge):
(12)
The obligatory word end at either 5 or 5.5 (rule 1) is the caesura, the most im- portant structuring element in the verse. The rationale behind the caesuraseems to be that a verse of the length of the dactylic hexameter is too long to
be perceived (and composed) at a stretch. The caesura structurally divides theverse into two, in such a way that (i) the second half of the verse is somewhat
larger than the first, (ii) the caesura does not fall between two feet (metra) butwithin a foot (the third one), and (iii) the rhythmical beginnings of the twocola are different: the first colon begins in a falling manner (beginning with ametrically marked element), whereas the second colon begins in a rising man-ner (see Van Raalte (1986: 70)). This threefold asymmetry contributes to theunity and coherence of the verse as a whole.
In the case of rule 2, we are dealing with what is commonly called the' buli diaeresis'. This phenomenon may be described as a desired (ratherthan obligatory) word end at 8 (i.e. after the fourth foot, which is realized as a
dactyl). Beekes (1972: 4) argues that the bucolic diaeresis must be seen as atendency resulting from the desire to end the verse in a truly dactylic manner.The coincidence of foot end with word end is avoided throughout the verse formetrical reasons (except at the beginning and the end of the verse: pos. 2 and10), but at 8 it is desired for the same reasons. It produces a rhythmically pleasing closing cadence, which was termed above (5.1.2) the clausula'.
The bridges at 3.5 and 7.5 (rules 3 and 4) may be explained in close connection with the diaeresis. Word end at 3.5 was avoided because it would involve in many cases a repetition of the clausula (-uu-u) of the line before,which would produce a monotonous and unaesthetic effect. As for Hermann's
bridge, word end at 7.5 is ill compatible with the diaeresis: word end both at7.5 and at 8 necessitates the occurrence of a monosyllable in 8, which is not
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 178/318
168 LINGUISTICS AND FORMULAS IN HOMER
easily realizable. And when the diaeresis is absent, word end at 7.5 would suggest, as Beekes notes, a 'false start' at the end of the verse (in comparison withthe right new start of the diaeresis). Moreover, as Van Raalte (1986: 98) notes,word end at 7.5 would involve the close (and hence unattractive) repetition ofthe rhythmical effect of the trochaic caesura in the third foot: ⋃ - ⋃ |5.5 ⋃ - ⋃7.5.
The rationale behind rule 5 seems to be the desire to avoid the suggestionof verse end. Any word end after a spondaic (bisyllabic) foot produces a verseend effect, which becomes stronger, and hence more to be avoided, near the
end of the verse. This explains that word end at 10 is downright forbidden,whereas at 8 it is merely strongly avoided (see also van Raalte 1986: 99,102).Rule 6 means in practice that a verse preferably does not end with a
monosyllabic word. This minor (in comparison with rules 1-5) rule is notwithout importance for the present study as per is often placed at the end of theline. More discussion of this is better in place in the following subsection(5.2.1.2).
A full picture of the bridges and preferred or obligatory word ends in thehexameter comprises mention of the frequent word end at position 2, and especially at 3 and 7. These word ends have no place in Beekes' set of rules, because,
according to Beekes, they have no direct rhythmical significance. Beekes (1972:3) contends that the fact that often 7 (and 3) are the place of syntactic breaksin the verse does not imply that these places have a structuring function froma metrical point of view. However, they are not wholly devoid of metrical interest. When the main caesura is absent, or only realized in a 'suboptimal' way,word end at 3 and 7 tends to gain in significance, both from a rhythmical andfrom a syntactic point of view. See further below.
5.2.1.2 On the notion of 'word-type'
The above account of the structure of the hexameter may suffice to copewith the first deficiency of O'Neill's approach, the lack of an explanatory foundation. We now go on to discuss a second disadvantage of his method.
In quantifying the occurrence and localization of word-types, it is usefulto have maximum clarity on the notion of 'word'; in other words, we have toknow what it is that we are counting. There are two points at which this question becomes urgent, one of which is formulaic in nature and the other linguistic. The latter is recognized by O'Neill, the former is not.
To begin with the former, sometimes a particular word is so firmly em
bedded in a stereotyped word-group ('formula') that the word-type to which it belongs has little importance in comparison with the 'word-type' formed by the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 179/318
PARTICIPLES Π : FORMULAS AND METRICS 169
formula as a whole. The result is that the localization of this particular word-token may differ sharply from the localization of the word-type found byO'Neill. Hoekstra (1965:22-4) gives some examples of this phenomenon. Theword isotheos ('godlike') occurs exclusively at 11, thereby violating Beekes' rule6, whereas the preferred positions of the metrical word-type to which it belongs(-⋃⋃-) are at 3 (verse-initial position) and 5 (before the penthemimeralcaesura). The explanation of the divergence is that isotheos exclusively occursin collocation with phôs ('man') to form a formula which is exclusively local
ized at 12 (isotheos phôs making a bucolic clausula). The localization of theword-string (formula) as a whole evidently takes precedence over the localization of its constitutive parts (the - u u - - word-type occurs in 12 and nowhereelse).34
Thus, in O'Neill's approach the distributional constraints upon particularwords which are due to formular language use are left out of account. Themetrical word-types studied by O'Neill may comprise more than one word.Once this is recognized, the localizational pattern of word-types in thehexameter will turn out to be even more exceptionless than appears fromO'Neill's tables.36
The second respect in which words may not be what they seem to be involves the presence of enclitic, phonetically less discrete words which may bementioned under the general heading 'prepositives and postpositives'. HereO'Neill does have a declared policy, but it turns out to be the wrong one. O'Neillinvariably treats postpositives as metrically separate from the words to whichthey adhere. This decision is made on the strength of lines as the following(1942: 109):
(13) oikad' ion sun nêusi te sêis kai sois hetaroisi
"Going home with your ships and your comrades." ( . 1,179).
(14) êdê gar kai deuro pot' êluthe dios Odusseus"For once Odysseus has been here." ( Il.3,205).
The enclitics here {te in (13) andpot'(e) in (14)), when they are taken togetherwith the preceding word, form a dactylic word ( - ⋃⋃) which fills the third foot.As this word-type is never localized this way (i.e. occurring in 6 and thus br idging the caesura), O'Neill concludes that te and pote, and hence postpositiveencUtics in general, have to be treated as phonetically discrete and rhythmically independent words. However, (13) and (14) are by no means equivalent, and,
hence, the class of enclitics is not uniform.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 180/318
170 LINGUISTICS AND FORMULAS IN HOMER
O'Neill's policy seems to be right in the case of (14) and similar examples.To treat deûo and pot'(e) as two separate words, thereby recognizing a realization of the T-caesura (albeit a 'sub-optimal' one), is a plausible account. In (13),on the other hand, the situation is different. The bond, both semantic and
prosodic, between te and the preceding word is much stronger than in the caseof pote and other particles (cf. Lat. -que). Accordingly, I have no difficulty withdenying (13) a caesura, or with admitting only a very weakly realized one.
The absence of the caesura in (13) is not so exceptional as O'Neill claims.
Actually, this verse conforms to a pattern which O'Neill does not recognize assuch. When the caesura is either absent or weakly realized, word end at 3 and7 tends to gain in rhythmical significance, yielding three rhythmical cola of increasing length. These cola tend to coincide with the syntactic cola of which theverse consists. This is what happens in the case of (13): this verse has a tripartite syntactic structure ('going home | with your ships | and your comrades')which coincides with a tripartite rhythmical structure:
(15) - ⋃ ⋃ - |⋃⋃-⋃⋃- |⋃⋃-⋃⋃--
Hexameters of this type are called in Kirk (1985:20) 'rising threefolders'. They play some role in the localization-pattern of 'participle + per'. Further discussion below (5.2.2 sub Π Bl)).
Thus we see that the notion of 'word' within the framework of the innermetric of the Homeric hexameter is elusive in two respects. Metrical word-types as the correct object of study and quantification may consist of more thanone 'word', both in the case of stereotyped word-groups and in the case of
postpositives. This critique of O'Neill's method is highly relevant for the present study, for per is an enclitic and 'participle + per' is used as a formulaconsisting of two or more 'words'. The bond between per and the preceding
participle or adjective is so strong that 'participle + per' should without moreado be considered one word. Once this decision is made, it is obvious that thedivergence from O'Neill's findings will be striking, for 'participle + pef makesa word-type which is entirely different from 'participle alone'.
The participle akhnumenos provides adequate illustration of this point. Akhnumenos is an instance of word-type - u u - , which, according to O'Neill'stable 15 (1942: 144) is preferably placed, in the Iliad, in position 3 (42%) or 5(37%). 18% of the cases is located in position 9 and only 13 % in position 11.
40
Now, there are 27 instances of akhnumenos and its inflected forms in the Iliad.
Of these only 3 ( = 11%) occur in position 3; 5 ( = 18.5%) occur in position 5,whereas 6 instances ( = 22%) occur in position 9 and no less than 13 instances
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 181/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 182/318
172 LINGUISTICS AND FORMULAS IN HOMERsemantics'. As has been stated, the best-known and best described examplehere is the noun-epithet formula. Odysseus, for instance, is called 'noble','many-counselled', 'much-suffering' and 'sprung from Zeus' purely in accordance with the metrical position which the name is to occupy (see Parry 1971:39). The descriptive content of the epithets is subservient to localization pur
poses.43
We now go on to discuss the localization of 'participle + per' in thehexameter. The issue at stake here is, among other things, the distribution of
the particles kai and, to a somewhat lesser degree, mala and empês in the participial phrase. I will show that these particles are very consistently used in orderto (i) adapt the form of participial phrases to the exigencies of the hexameter,and (ii) make rhythmically untractable combinations of 'participle/adjective +
pef fit into the metric profile of the verse. Kai, mala and empês thus constitutean important case of the typically formular 'redundant semantics'. Just as theepithet in noun-epithet formulas, their use is in service of automatic andsmooth versification. This means that the proper meaning of the particles inthe language is subservient to this purpose. Thus, in many cases it does not makesense to look for a justification of a given instance of kai in the usual terms of'strengthening', for the occurrence or non-occurrence of the particle is decidedon entirely different grounds. I shall have occasion to show that what appliesto kai, mala and empês applies to less regularly occurring elements in the participial phrase as well.
The meaning of epithets in noun-epithet formular systems has to be in accordance with the character of the hero in question. Otherwise the epithet cannot properly perform its automatic, word-type changing function. The sameapplies to the particles in the participial phrase when we view them from the
point of view of 'redundant semantics'. Their meaning has to be 'compatible'
with that of per if they are to perform their versifying function with any success.Since in ordinary language the occurrence of the particles in question in participial phrases has to be stated in terms of strengthening, this 'compatibility'comes as a matter of course. Kai is an inclusive focus particle ('also', 'even'),whose affinities with scalarity and concession need no further commentanymore. And mala ('very') is in ordinary language a very natural strengthenerof the superlativeness of the scalar-concessive 'participle + per' Thus, we maysay that in using kai and mala for versifying, word-type changing purposes, epicdiction aptly exploits the normal semantic possibilities of these particles. As forempês ('anyway'), this particle properly belongs to the main clause, rather than
to the participial phrase. Its occurrence has somewhat more semantic conse-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 183/318
PARTICIPLES Π : FORMULAS AND METRICS 173
quences than that of kai and mala, and in ordinary language its use cannot becharacterized in terms of strengthening. See further below (sub I.B1), I.B2)).
In the presentation below of the localization of 'participle + per', theword-type of the participle or adjective/substantive in question is taken as thestarting-point each time. The basic word-type is normally automatically extended with a short syllable ( = per) in the case of 'participle' (i.e. synthetic participles), and with ⋃⋃- (per eon) or ⋃⋃-⋃ (per eontos l-ti etc.) in the case of'adjective' or 'substantive' (i.e. analytic participles). I start with the first group.
I. Synthetic participles
A. Middle participles
Middle participles are always inflected as o-stem nouns. Consequently,inflection here does not result in a change in the word-type formed by the
phrase as a whole.45
Al) Word-type -⋃⋃ - . The most common localization here is at verse end,as a bucolic clausula. This is the formular use of 'participle + per 9 recognized
by Parry (1971: 311-12):
(16) -⋃⋃-⋃⋃-⋃⋃-⋃⋃8akhnumenos per
hiemenos peroutamenoi perteiromenos perkêdomenon perollumenoi peressumenos perkhôomenos per
When the participle has an initial vowel, it is never preceded by kai. Thisconstraint on the distribution of kai has an easy metrical explanation, just asthe positive occurrence of the particle elsewhere: kai, being shortened by theinitial vowel ('epic correption'), extends the metrical length of the participial
phrase by a short syllable. The word-type which would result from thislengthening (⋃-⋃⋃- - ) cannot be placed at the end of the verse, because itwould result in word end in 7.5, i.e. violation of Hermann's Law (see rule 3 in(11) above). Kai before a consonant, on the other hand, is unobjectionable; ityields the word-type — ⋃⋃ — (for example, kai kêdomenon per), which can
be conveniently placed after 7.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 184/318
174 LINGUISTICS AND FORMULAS IN HOMER
The localization of word-type - ⋃⋃ - - at the end of the verse allows, interestingly, for the substitution of a monosyllabic substantive, kêr ('heart') for
per. This substitution occurs in the case of the two participles in the Hst of (16)which denote emotions, akhnumenos ('grieved') and khôomenos ('angry'). Kêr
functions as a semantically redundant 'accusative of respect' ('grieved/angry inhis heart').46 It is obvious that this substitution possibility greatly augments theformular usefulness of the localization of akhnumenos/khôomenos (..) at verseend: whenever a scalar-concessive reading is out of place, one simply uses
akhnumenos kêr, thereby eliminating the meaning of per but retaining itssound.The second common localization of the akhnumenos per word-and for
mula-type is at 9.5:
(17) -⋃⋃-⋃⋃-⋃|⋃ 6akhnumenos per9.5⋃ —essumenos peroutamenoi per
There are 13 instances of this localization-type (8 times akhnumenos per ) , all
of which begin, significantly, with a vowel. Of these 13 instances, no less than12 are preceded by kai. Within the metrical framework, this constant occurrence explains itself: akhnumenos per as localized in (17) leaves a short syllableafter the trochaic caesura. This short syllable can be very conveniently occupied
by kai, which in lengthening the basic word-type to ⋃ - ⋃⋃ - yields the word-type which is discussed under A2) below (⋃-⋃⋃-⋃). Thus, just as the insertion of kai before akhnumenos per in 12 is forbidden on metrical grounds, so itis highly preferable, if not mandatory, in case of the localization in 9.5, equally on metrical grounds.
The composite word-type exemplified by kai akhnumenos per localized in
9.5 leaves open ⋃ — as metrical space at the end of the line. This space is often,characteristically, occupied by a word which may be conceived of as an argument of the participle seen as a predicate (in 4.3.2.3 we saw that adjectival circumstantial participles like akhnumenos may be seen as equivalent to the
predicate of an ascriptive sentence). Good examples of this phenomenon arekai akhnumenos per hetairou ('though grieved about his comrade'),48 and inthe case of essumenos per. kai essumenos per hodoio/Arêos ('though eager forthe road/battle'). Another formular and stereotyped filling of the ⋃ — finalslot is anagkei ('perforce'). This adverbial element,50 which syntactically
belongs to the main clause, is compatible with the scalar-concessive meaning
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 185/318
PARTICIPLES II: FORMULAS AND METRICS 175
of the participle: one has to do something while being forced by the circumstances. So one does the thing, grieved as one is. An example:
(18) - ⋃ ⋃ Argeioi de kai akhnumenoi per anankêinêôn êmunonto."And grieved though they were, the Argives were compelled to fightfor their very ships." ( Il. 12,178). Cf. . 14,128; 15,133.
Only in two cases, interestingly in close proximity of one another (Od. 4,
549,553), is the slot between 9.5 and 12 occupied by a verbal form which continues the syntax of the sentence.
The reason for the preferred filling of the final ⋃ — slot is obvious: byfilling it with anagkêi or a stereotyped argument of the participle, one acquiresone long scalar-concessive phrase running from the trochaic caesura to the endof the line. Such a phrase is highly convenient in a formulaic diction in that itmakes a useful T2-expression. The poet localizing akhnumenos per in 9.5 neednot worry about the remaining open metrical space at the end of the line.
The third possible localization of akhnumenos per and its type is in 5.5,
just before the trochaic caesura:(19) - ⋃⋃ 2akhnumenos per5.5 ⋃-⋃⋃-⋃⋃ - -
teiromenoi peressumenon per
Here there is, for once, no compelling metrical reason for either the presenceor the absence of kai before an initial vowel: word end ( = word beginning) in2 is slightly preferred to word end in 1.5, but the difference is not significant.Consequently, kai just may or may not occur in this case of localization. Moreimportant and more interesting in the present connection, however, is the following. Localization in 5.5 leaves the metrical space of a dactyl (-uu) open atthe beginning of the line. This is exactly the space needed for the collocationof kai and mala. Before an initial consonant this collocation may be conveniently inserted. This is what happens in the following example:
(20) hêmeis men Danaoisi makhesometh', authi menonteskai mala teiromenoi per, anagkaiê gar epeigel"We will stand our ground and fight the Danaans here, exhausted
as we are - we have no choice in the matter." ( Il. 6, 85).
We have here a fine example of deliberate localization and redundant semantics: out of the ordinary language expression teiromenoi per, which is preferab-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 186/318
176 LINGUISTICS AND FORMULAS IN HOMER
ly used as a bucolic clausula, is made, by means of the combination kai mala,
an entire Τ ι-expression which fills the first part of the line.51 But the epic poetcan go even further. Consider:
(21) (Odysseus to his host Alcinous: 'please let me eat my supper, fornothing is so shameful as a man's stomach:)'
-⋃⋃hê ekeleusen heo mnêsasthai anagkêikai mala teiromenon, kai eni phresi penthos ekhonta"(However afflicted he may be and sick at heart), it calls for attentionso loudly that he is bound to obey it." {Od 7,218).
The mini-scale distributional pattern of per in (20) and (21) can be explainedwithin the framework of localization. The participial phrase without per in (21)has the same meaning as the participial phrase with per in (20). The differenceis formular and localizational. In composing the line Od, 7,218, the phrase kaimala teiromenon per as a pre-caesural (T1-) expression was of no avail to the poet, as the caesura in the verse in question was to be realized as P. So he simply skipped per, changing a Τ ι-phrase into a Pi-phrase. We have here the ultimate example of the extent of the influence of localization and versification:the central semantic element in the whole, per, is made subservient to the fluctuation between Ρ and Τ caesuras, which is, in its turn, frequently put to for-mular ends.52 Thus we see that by a process of systematic deletion and additionof particles, a poet may make a Τι-formula out of a bucolic clausula, and a Pi-formula out of a Τι-formula. Even per itself is made subservient to the
phenomenon of redundant semantics.
A2) Word-type ⋃-⋃⋃-. This is the second basic word-type we discuss.Localization in 12 is impossible, because of Hermann's Bridge (see rule 3 in(11) above), the same reason why kai was impossible in (16). So the remaininglocalizations are 9.5 and 5.5:
(22) -⋃⋃-⋃⋃-⋃ | 5.5sepeigomenos p 9.5⋃ —lilaiomenon peroïomenos perhimeiromenos per
(23)) - ⋃ 1.5epeigomenên per5.5 |⋃-⋃⋃-⋃⋃ —
This word-type and localization-pattern does not, and need not, involve the use
of word-type changing particles. The present word-type localized in 9.5 is in allrespects equivalent to akhnumenos per and its type localized in 9.5 and
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 187/318
PARTICIPLES Π : FORMULAS AND METRICS 177
preceded by kai (see (17) above). Epeigomenos occupies the metrical slot towhich (kai) essumenos has been adapted. In the case of essumenos, too, theremaining slot (⋃--) at the end of the Hne is filled by either anankêi or arguments of the participle. Epeigomenos per (hodoio/Arêos) and kai essumenos per(hodoio/Arêos) are simply the prosodic complements of one another, the one
beginning with a vowel and the other with a consonant. The lexical differenceis subordinate to the formular use which is made of the phrases.
A3) Word-types ⋃⋃ - uu and -- ⋃⋃. These word-types, which are formed by middle perfect participles, are different from the two previous ones ( - ⋃⋃ -
and ⋃ - ⋃⋃ - ) in that they end in a pendant (dactylic) manner, i.e. on a doubleshort, metrically unmarked element. This has consequences for the use of perin connection with this participial word-type, for per cannot be placed after the
participle: its initial consonant would cause the final syllable of the participleto be long by position. This is impossible, for single-short sequences ('cretics':- ⋃ - ) cannot occur in the hexameter. So per has to be placed before the participle. But per, as an enclitic, has to stand after something. Here mala has itsrole; it serves as a phonetic substrate for per. This results in the following
localizations (either in 8 or in 10):(24) - ⋃⋃ - 3mala per kekholômenons - ⋃⋃ - -
kekhrêmenos
(25) - ⋃⋃ - ⋃⋃ - smala per kekholômenoso - -
In the case of (24), kai may be added ad libitum, yielding a rhythmically convenient congestion of particles (kai mala per), see Od. 14,155. This collocationis once ( Il. 1, 217) separated from the participle by a semantically empty substantive (thumôi ' in his heart') which is equivalent in meaning and function tokêr in akhnumenos kêr (see above). The result is an extended participial phraserunning from the beginning of the line to the bucolic diaeresis:
(26) kai mala per thumôi kekholômenon 8 - ⋃ ⋃ - -
"However much you are angry in your heart."
In the case of (25), which occurs only once, the remaining final slot after 10 (--) is filled by an intensifying adverb, ainôs ('terribly'), which makes out of thelocalization in 10 a P2-expression.55
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 188/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 189/318
PARTICIPLES Π : FORMULAS AND METRICS 179
(30) kai mala per phileonte5.5
b) The second strategy to handle the word-type consists in ' per +
empês' after the participle instead of mala per before it. Localization is 100%in 12:
(31) 7 pinonta per empês12
khateontikhateousi
The memaôta-forms of (28) do not occur here. The forms of khateô ('need'),on the other hand, occur in both localization patterns. The fact that two metrically and prosodically equivalent phrases (mala per khateontin12 and khateonti
per empê12) exist side by side, as well as the fact that memaôti can only be substituted in the first case, may be explained by the fact that there is a semanticdifference between mala and empês. Mala is superlative and is, accordingly, illcompatible with participles which do not allow of superlativeness. Empês, onthe other hand, is concessive. Furthermore, mala properly belongs to the participial phrase, whereas empês belongs to the main clause, just like anankêi
above. This difference in meaning seems to adequately account for the existence of two rhythmically and semantically equivalent formulas at the sametime.60
B2) Word-type...] i.e. word-types ending as the previous one, butlonger (beginning earlier in the verse). Here per empês after the participle isthe standard. The result of localization in 12 is always a T2 or P2-expression.Sometimes kai is inserted to fill the remaining space between the beginning ofthe participial phrase and the P-caesura:
(32) 5.5epikrateousi per empêsapekhthomenoisi5oligêpeleousa5kai tethnêôta
II. Analytic participles
Classification here is based upon the word-type of the adjective or substantive. The treatment per basic word-type has to be divided into two, according to the case ending of the participle eon: in the case of nominative singular
we have a rhythmical sequence ending on...] (...]per eon), and in the caseof all other cases we have...] (...] per eontos etc.). The latter is potential-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 190/318
180 LINGUISTICS AND FORMULAS IN HOMER
ly ideal for localization in 12 or 5.5 before the T-caesura, whereas the formeris not.
A) Adjectives of word-type
A1) per eon). Localization of analytic participles of thisword-type in 7 ( = beginning in 3.5) would involve a violation of Meyer's FirstLaw (see rule 4 in (11) above) and is therefore precluded. Hence localizationeither in 5 or 9:
(33) 1.5geron per eon
(34) 5.5ligus per eon9thoosgerôn
Notice that in these localization patterns the word end after per occurs in a position where it is normally avoided: 3.5 in (33) (Meyer's First Law) and 7.5 in(34) (Hermann's Bridge). That these metrical rules are consistently violated isan index of the coherence of the participial phrase: apparently the participleeon was not felt as a phonetically discrete word. Thus word end in 3.5 and 7.5at the boundary of the participial phrase is consistently avoided, while withinthe participial phrase it is consistently allowed.
Localization in 9, as in (34), leaves open at the end of the line. Thisslot is invariably occupied by a substantive belonging to the participial phrase,turning the adjective from an adjectival predicate into an adjective modifyinga substantival predicate. In the case of ligus, we have ligus per eôn agorêtês ('shrillspeaker') and in the case of thoos and gerôn we have thoos/gerôn per eon
polemistês ('nimble/ old warrior').64 The substantival extension yields a par-
ticipial phrase with the convenient length of a T2-expression.A2) per eontos). Localization at verse end (12) is metrical
ly objectionable as the participial phrase would begin in 7.5, thereby violatingHermann's bridge (see rule 3 above). Hence localization in 5.5 and 9.5,analogously to the non-oblique version in la). Notice that there is the sameseeming violation of a metrical rule as in (33) and (34) (word end in 3.5 and7.5):
(35) 1.5lithos per eousa5.5
ernen per eousannothon per eonta
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 191/318
PARTICIPLES Π : FORMULAS AND METRICS 181
megan per
philon
(36) neon per eonta
nekun per
Participial phrases of the basic word-type are never preceded by kai. Thereason is clear: in case of localization in 9/9.5 the phrase begins at the trochaiccaesura, which simply precludes kai, and in the case of localization in 5/5.5 km
can only occur before an initial vowel (otherwise a hexametrically impossible'cretic' would result: ). But the adjectives in (33) and (35) begin with a
consonant (except emên in (35)). The eon/eontos-fluctuation does not result intwo separate localizations of the phrase, as both 5/5.5 and 9/9.5 are normal
places for word end. This is different in the following word-types.
B) Adjectives of word-type
B1) per eon). The most common localization by far is in7, the phrase starting in 3:
(37) 3athos per eôn7gumnosl-onkhlôros
zôoskrateroseuruskuneosautoskseinos
This localization-type, which divides the verse into three, should be discussedin connection with the tripartite hexameter discussed above (see (15) in 5.2.1.2above), the participial phrase occupying the second colon bridging the caesura.
Now I do not want to suggest that the hexameters exemplified by (37) do nothave a caesura at all; actually they have one, between per and eon. But we have
just seen (in (33)-(36)) that 'word-boundary' between per and eon is not verystrong. Accordingly, the caesura in the cases exemplified by (37) is certainlynot strong ('prototypical') either. Furthermore, the combined rhythmical andsyntactic importance of positions 3 and 7 in the hexameters exemplified by (37)
is strongly reminiscent of the 'rising threefolder' The rising three-fold structure seems to be the right (rhythmical) explanation for the near absence of kai
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 192/318
182 LINGUISTICS AND FORMULAS IN HOMER
before the participial phrases exemplified by (37): the addition of kai makes adactylic ( = falling) beginning of the participial phrase which spoils the risingrhythmical movement.
There are two instances of localization of the adjectival word-typein 5. It goes almost without saying that the remaining open position at the
beginning of the Une is filled by kai:
(38) kai krateros per eons pezos
B2) per eontos). This is the inflected pendant of theword-type just discussed. It cannot begin in 3, as in (37): its end would fall in7.5 with a serious violation of Hermann's Law. The preferred localization is 12,after 7:
(39) 7alokhôi per eousêikranaes -ses
polees eontes pollôn -tondeinon -ta
pleonas -taskraterô -te
pinutê eousa
Kai never precedes here, no doubt because it would leave too little open space between the caesura and the beginning of the phrase to be workable from the point of view of versification (see also the localization of C2) below).However, in the case of the second localization possibility of the type in ques
tion, 5.5, before T, kai
invariably precedes, making out of the phrases exemplified in (39) a Τι-expression (for this principle compare (28) and (30)above):
g(40) kai nekuos per eontos5.5.5
hiketên -ta pêôi -tikhalepon -ta
polioi -tes
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 193/318
PARTICIPLES Π : FORMULAS AND METRICS 183
Note that the instances exemplified by (40) may also be treated as the in
flected form of the cases under (38): the eôn/eontos-alternance neatly exploitsthe fluctuation between the P- and Τ caesura (see also (20) and (21) above).
C) Adjectives of word-type
C1) per eôn). Localization in 7, always:
(41) 2allodapos per eôn7
kheiroteroshoploteros
Kai just may or may not precede here, as there is no compelling reason why the phrase should begin in 2 rather than in 1.5 or vice versa (see also on (19), thelocalization of akhnumenos per in 5.5).
C2) per eontos). Localization either in 5.5 or in 12.In the first case the phrase is a ready-made T1-expression:
(42) pauroterous per eontass.5
khruseiê eousatoksophorôi -sêiôkutaton eonta
In the second case (localization at verse end), the poet only needs to putkai before an initial vowel to get a T2-expression; kai always occurs:
(43) 5.5kai iphthimôi per eonti-mon -ta
orkhêstên
ouk agathonD) Adjectives of longer word-types: and
Dl) per eôn). Localization in 7:
(44) 1smesaipolios per eôn7
D2) per eontos). Localization always in 12, yielding either a T2- or a P2-expression:
(45) 5.5epistamenô per eontiminunthadion -ta
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 194/318
184 LINGUISTICS AND FORMULAS IN HOMER
sdourikteten per eousan
Kai naturally does not occur here; the caesura precludes any expansion of the phrase.
E) Adjectives of word-type and
These word-types can be taken together. They differ sharply from the previous ones A) through D). They either are or end dactylic, ending on a metri
cally non-marked element, while the word-types just discussedand ) end on a metrically marked element. In the case of theseword-types, the direct addition of per eônleontos guarantees a smooth rhythmical continuation. In the case of and on the other hand, it does not:it would yield a sequence of no less than four short syllables, which is impossible in the hexameter. The solution is the insertion of mal' (a) between per andeonta, which makes of per a syllable which is metrically long by position: permal' eonta. Only oblique forms occur; localization is always in 12:
(46) eurea per mal' eonta
(47) epiphrona per mal' eontaapênea
Kai may do much good here: in (46) it is virtually obligatory, as the metricalspace between the beginning of the phrase and the caesura is too insignificantto be used for anything else. In the case of (47), we have a ready-made T2-ex-
pression. The addition of kai may, if need be, enlarge the phrase to the lengthof a P2-expression. This is the case in ex. (43) in 4.4.1 {kai epiphrona per maV
eonta). Notice that the insertion of per mal' before an adjective is sometimes, too,
the apposite strategy in the case of scalar superlatives in a non-factual context(see 3.1.1), for example poluphrona per mal' aeisai (Od. 14,464). The final slot( ) which is occupied by eontos/-ta in (46)-(47) is here occupied by anotherelement which continues the syntax of the sentence. We may say that from the
point of view of the versification of scalar superlatives, the participial form eontosl-ta is an element belonging to the formular system: when there is no finalword of the form available, the poet can always resort to eontosl-ta, there by turning his scalar superlative into a circumstantial participial phrase with theform of a T2-expression. In 4.4.1 above I have shown that this is linguistically
possible.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 195/318
PARTICIPLES II: FORMULAS AND METRICS 185
F) Adjectives of word-type
This word-type ends in a trochaic manner. Just as in the case of the previous word-type, per eontos cannot be added without more ado; this time itwould result in the sequence of three short syllables (tribrachys). The strategyof the previous word-type is metrically objectionable, as per mal' eontos after
would lengthen the final short syllable of the adjective, with the result12. This word-end in 8 after a spondaic fourth foot is awkwardand should be avoided (see rule 5 in (11) above). A rhythmically much more
pleasant solution is mala per before the adjective:
(48) - mala per polumuthon eonta poluidrin eousan
The result is a ready-made P2-expression with an internal structure like that ofmala per memaôta (makhesthai), see I.B1) a) above with ex. (29).
We have now arrived at the end of the discussion of the regular andanalogically productive strategies in the localization of 'participle + per 9. A few
cases remain, which are isolated and/or irregular. They do not conform to a productive pattern. An isolated but recurrent case is the formula kaka per paskhontes localized in 9.5, to be discussed in 5.3.2 below. Comparable are ek-honti per algea thumôi and phugonti per aipun olethron, discussed in 4.3.4.
The isolated and unique cases are the following. They are characterizedeither by manipulation with kai or by a substantive in the phrase, which servesas an argument of the participle:
(49) kai per polla pathonta5.5 (Od. 7, 224)
(50) skai polla per athlêsanta ( Il. 15, 30)
(51) 5kai kêdea perpepathuiêi (Od. 17, 555)
(52) kai tethnéota pers ( . 24, 20)
(53) 5.5kai en thanatoio per aisêi (Il. 24, 428)
(54) spotamoio per ekgegaôti ( . 21,185)
(55) 1.5kai ergomenê mala p r7 Il. 17,571)
Ex. (49) is the only instance of the direct collocation of kai + per in Homer.After Homer it has become the normal expression.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 196/318
186 LINGUISTICS AND FORMULAS IN HOMER
The above discussion of the localization of 'participle + per 9 has provedthat the distribution of kai, mala and empês in participial phrases can be ex
plained within the framework of the formular system constituted by 'participle+ per'. By way of summary, we may distinguish three factors, one for each particle:
1) A participial phrase may leave open metrical space between its beginning and the caesura or the beginning of the line or another desired word end.Here kai (once in combination with mala: see exx. (20)-(21)) is inserted. Itsfunction is backward extension of the participial phrase.
2) A participial phrase may leave open metrical space between its end andthe end of the Hne; it needs forward extension. Here empês comes into the picture, along with some semantically compatible and/or innocuous substantives(anagkêi, hetairou).
3) A basic word-type, either of an adjective or of a synthetic participle,together with per (eontos) may yield a sequence which does not fit into themetric profile of the verse. Here mala has its job; together with per it formsconfigurations {mala per, per mal*) which make the use of the word-type in
question as the basis of a participial phrase possible.
As for exx. (49)-(55), these cases are irregular as to their form and internal structure. As to their meaning they are normal instances of the concessive
participial phrase. In being isolated cases which do not conform to any pattern,formulaic or linguistic, they seem to be ad hoc solutions for versification
problems encountered by the poet.
5.3 Formulas and meaning: the integration of formulas
One of the characteristic features of the formular use of language is thatoften the need under given circumstances to use a particular phrase (formula),takes precedence over factors that are, in a non-formular context, less favorableto that use. The poet's need to use the phrase is a direct consequence of theusefulness of the phrase as a formula, viz. as a standardized 'buiding-block'.The degree to which a formula fits into its context I state in terms of integration. When dealing with the integration of a formula, or the degree to whichit is integrated, we compare fully integrated instances of the formula to imperfectly integrated ones. Accordingly, in terms of the levels of recurrence discussed in 5.1.3 above, we are concerned here with the verbatim repetition of
formulas.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 197/318
PARTICIPLES Π : FORMULAS AND METRICS 187
5.3.1 Levels of integration
The integration of formulas has a number of different aspects. In otherwords, a formula may be less than perfectly integrated in more than one way,on various levels. One of these levels is phonetics and prosody. Here, too, Parryhas done pioneering work. In his second French thesis (1928b), he showed thathiatus in Homer (the direct contact between word-final and word-initialvowels, which is avoided in all Greek poetry and cultivated prose) as well asother metrical irregularities, are not due to some set of 'rules', on account of
which the irregularity was allowed in some places of the hexameter and 'illicit'in others. Nor did hiatus occur because it was allowed anyway in Homer, in contrast with later authors.
Parry showed that hiatus and brevis in longo (a prosodically short syllablein a metrically long position) was due to, and a consequence of, the modification and juxtaposition of formulas.68 'Modification' is an internal change in theformula, flexion, for instance, which may cause hiatus. It is with 'juxtaposition'that we are here mainly concerned. Juxtaposition may cause a case of hiatus or
brevis in longo between two formulas. Juxtaposition thus involves the use
which is made of formulas within a given context, and it should, accordingly, bediscussed in connection with the degree of integration of a formula in its context.
Two examples are sufficient to illustrate the less than perfect phonetic integration of a formula in its context (both are given by Parry, 1971: 203; 215):
(56) entha kathezet' epeita Odussêos philos huios
"And there the son of Odysseus sat down." (Od. 16,48).
There is hiatus between the final vowel of epeita and the initial vowel ofOdussêos. This is because Odussêos philos huios, a standard T2-formula forTelemachus (its natural locus being between the trochaic caesura of the thirdfoot and the end of the line, see 5.2.1.1) is used in a phonetic context in whichit does not really fit. It occurs in a context which is ideally suited for a formulaof the same metrical shape but beginning with a consonant. Such a formula occurs in
(57) entha kathezet' epeita polutlas dios Odusseus."And there the noble and stalwart Odysseus sat down" (Od. 19,102).
Here the noun-epithet formula is fitting, and the phonetic continuity of the
verse ('synhaphea') is not interrupted, as in (56).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 198/318
188 LINGUISTICS AND FORMULAS IN HOMER
In the second example there is brevis in longo before a P2-formula (locus between the penthemimeral caesura and the end of the Une, see 5.2.1.1). Theformula has already been cited, in 4.4.1:
(58) tosson apo ptolios hosson te gegône boêsas
"Within call of the city" [lit: "so far away from the city as he [unspec]can make himself audible by crying"]. (Od. 6,294).
The hosson te gegône boêsas-formula has to be preceded by a naturally long syll
able if it is to be fuUy integrated from the point of view of phonetics and prosody.This is the case in
(59) all' hote tosson apên hosson te gegône boêsas"But when he had come within caU of the shore." (Od. 5,400).
The metrical and prosodical irregularities in (56) and (58) should not betreated as legitimate' in any way. Rather, they are inevitable. The poet com
posing ex. (56) or (58) did not, in allowing the metrical irregularity, exploit somelicense or suffer from a sudden lapse of concentration; his need to use this par
ticular P2/T2-formula in this particular context simply took precedence over hisconstant desire to avoid hiatus and brevis in longo and to produce metricallyand prosodically perfect hexameters. The result is a formula which is not fullyintegrated in its metrical and prosodic environment. But in oral performanceand composition this is a thing one has to live with.
An entirely different kind of integration involves the appropriateness ofa phrase in a given context as to its content. Here metrics and prosody do not
play a role in the less than perfect integration. What matters is that what theformula says is not (entirely) in accordance with the context. The context hereis not to be conceived of as a phonetic continuum in which a phonetic discon
tinuity ('hiatus') has to be aUowed, but as a stretch of discourse in which whatis said is adequate and relevant. Imperfect integration on this level involves information that is less appropriate in the context, but which is nonetheless given
because it is part of a formula. The phenomenon has been caUed in Combel-lack (1965) the 'formular illogicaUty'. Combellack's best example of the
phenomenon (1965: 53) is the one cited here.In book 11 of the Odyssey, the everlasting punishment of Tantalus is
described. Tantalus stands in a pond with the water to his lips. Whenever hestoops to drink, the water recedes and vanishes in the ground. Over his head
fruit is hanging, which is blown away whenever he reaches for it. The fruit isdescribed as follows:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 199/318
PARTICIPLES Π : FORMULAS AND METRICS 189
(60) onkhnai kai rhoiai kai mêleai aglaokarpoi
sukeai te glukerai kai elaiai têlethoôsai
"Pears and pomegranates, glossy apples, sweet figs and flourishingolives." {Od. 11, 589-90).
One might want to argue that Homer did not know that olives on the tree areunedible, just as one might want to argue that hiatus is allowed in Greek epic
poetry. But a better way to account for the olives, which are irrelevant in thecontext of (60), is treating the two hexameters as a standard formular enumeration of 'fruit', whose integration in the context is less than perfect because oneof the kinds of fruit is out of place in the context. The same enumeration can
be found elsewhere in the Odyssey (7, 115-6), where the description of a garden provides a suitable environment. The need to use this standard formularenumeration of fruit trees has proved stronger than the desire to achieve factual correctness.
5.3.2 Prototypicality and the semantic integration of formulas
In the previous subsection I discussed two possible consequences of theformular use of a phrase. On the level of phonetics and prosody, the less than
perfect integration of a formula results in a discontinuity of the 'synhaphea'(phonetic continuity) of the verse. On the level of content, it results in a lessrelevant piece of information in a given context. In the present subsection, weare concerned with the level of integration that Hes in between, the level ofsemantics.
In 5.11 proposed a functional approach to the Homeric formula. I arguedthat phrases function as formulas when they are conducive to the poet 's fillingthe metrical space of the hexameter in a systematic way. Phrases are not to be
called formulas because of what they are but because of what their function is.Accordingly, there is no qualitative difference between formular and non-for-mular language use. The implication of this account of the formula is that thedegree of semantic integration of a formula has to be stated in normal linguistic terms. It is Prototype Theory, introduced in 1.4, that provides the terminology here.
When a formula is used under the conditions (pragmatic and/or syntac-tico-semantic) for which it was originally devised, we may say that it is a perfectly integrated formula. In linguistic terms, this means that it is a prototypical
instance of the phrase in question. A fully integrated formula is
a phrase whichis used in accordance with the semantic potential it possesses. And a phrase
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 200/318
190 LINGUISTICS AND FORMULAS IN HOMER
that is used in accordance with its semantic potential has all the prototypical properties of the expression-type to which it belongs. Now, in Homer, phrasesmay be less than prototypical (or even peripheral), precisely because they areused as formulas. The usefulness of a given phrase as a standard building-blockin the versification may imply its being used under semantically less than optimal circumstances. The formular approach to less than prototypical instancesof a given expression-type focuses on how the phrase in question comes to beused, whereas in the linguistic approach we merely describe it by confronting
it with the more prototypical instances.The joint operation of prototypicality/peripherality and semantic formular integration seems to be the right approach to a number of less than
prototypical instances of 'participle + per'. In the following two examples, theanalysis of Hiemenos per in a formular framework accounts for the less than
prototypical status of the phrase. In being only implicitly concessive, the participial phrases are reminiscent of exx. (35)-(37) in 4.3.4 above:
(61) (The Trojan Cleitus is shot by Teucer:)takha d'autôi
êlthe kakon, to hoi ou tis erukaken hiemenôn per"His punishment was swift, and one from which no zealous friendcould save him." ( Il.. 15,450 = 17,292).
(62) (A formular line from the Odyssey:)
α ll' Oduseus kateruke kai eskhethen hiemenô/ên/on/ous per
"But Odysseus held them/her/us back and checked their/her/ourimpetuous movement." (Od. 4,284; 16,430; 21,129; 22,409).
The participial phrases in these examples are far from prototypical. In (61), the
participial phrase has to be taken as a so-called 'partitive genitive' to ou tis: 'noone out of those who were zealous (could save him)'. The usual concessive meaning of the participle is, of course, not far away, as appears from the paraphrase'His friends could not save him, however much they wanted to do so'. This
paraphrase will have had considerable 'psychological' reality (see 4.3.4) in the poet's mind. But the actual wording, in which 'participle + pef plays, like all partitive constituents, a restrictive role, is very unusual. The restrictive partitive function is very different from the normal detached status of 'participle +
per'.The participial phrase in (62) is equally peripheral from the point of view
of the semantics of 'participle + pef. The concession Hes, of course, in the factthat the people in question, who are held back by Odysseus, are unable to per-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 201/318
PARTICIPLES Π : FORMULAS AND METRICS 191
form some action in spite of their eagerness. However, the participial phrasecan only have its normal scalar-concessive function when the main clause towhich it is attached is expressed as a passive sentence without an overt agent:'They were restrained in their movement, in spite of their eagerness'. But themain clause as it is expressed in (62) is transitive; it refers to a state of affairswith a clear agent: 'Odysseus held them back'. And with respect to this activesentence, the concessive participial phrase cannot have its usual detached
status: the paraphrase 'Odysseus held them back and checked them, (and he
did so) in spite of their eagerness' is not the intended meaning of (62). Now one could stop the discussion here, at the point were we can say that
the concession in (61)-(62) is only implicit, (61)-(62) simply being peripheral
instances of their kind. But it is possible, I think, to go further and explain whythe participles in (61)-(62) are used at all. Hiemenos per placed at the end ofthe line, as a 'bucolic clausula' (see 5.2.1.1), conforms to the very commonlocalization pattern discussed in 5.2.2 sub LA1. This pattern clearly has formular (functional) interest, as it enables the poet to fill a rhythmically significant part of the hexameter in an automatic way. The usefulness of'participle + pef is further enhanced by the fact that, as an optional constituent,it may be omitted without more ado, when the metrical space after the bucolicdiaeresis is to be filled with other linguistic material.
The formular usefulness sufficiently accounts for the less than prototypical status of the participles in (61)-(62). In these examples, it was more important for the poet to use hiemenos per anyway than to use it in its proper function,as an instance of 'participle + per', viz. to add salience to the main clause.
Hiemenos per in (61)-(62) does not add salience to its main clause. The sufficient condition for its use is its usefulness as a formula, and the necessary condition is the fact that there is after all concession present in the context, albeit
implicit and 'psychological'.I conclude the discussion of (61)-(62) with two examples in which the
reasons to use hiemenos per as a formula coincide with the reasons to use it asan instance of 'participle + per', Consequently, both instances are perfectly integrated formulas and at the same time prototypical instances of their expression-type:
(63) (From the proem to the Odyssey:)
all' oud' hôs hetarous errusato, hiemenos per"But he failed to save his comrades, in spite of all his efforts." (Od.
1,6).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 202/318
192 LINGUISTICS AND FORMULAS IN HOMER
(64) (From the description of the battle over Patroclus' body:)ôsan deproteroi Trôes helikôpas Akhaiousnekron de prolipontes, hupetresan,, oude tin' autônTrôes huperthumoi helon egkhesin hiemenoi per.
"At first the Trojans flung back the bright-eyed Achaeans, whoabandoned the corpse and gave ground before them. Even so thehaughty Trojans did not succeed in killing any of them with theirspears, for all the pains they took." (R 17, 276).
As the second example of the imperfect semantic integration of formulas,I give one of the examples of modal embeddedness that was already discussedin 4.3.4 above:
(65) mê de houtôs, agathos per eên, theoeikel'Akhilleuklepte noôi, epei ou pareleuseai oude me peiseis"Do not imagine, although you are a great man, that you can trickme into that. I am not going to be cajoled by you." ( Il..1,131).
The description of this and similar examples as less than prototypical instances
of the expression-type 'participle + per' was presented in 4.3.4 above. We nowdiscuss the phrase as a less than perfectly integrated formula. For a full understanding of (65), we have to notice that agathos per eon occurs elsewhere in
book 1 of the Iliad and that the whole line occurs in book 19. Consider:
(66) (Nestor to Agamemnon:)mête su tond', agathos per eôn apohaireo kourênall' ea, hôs hoi prôta dosan geras huies Akhaiôn."Agamemnon, forget the privilege of your rank, and do not rob himof the girl." (R. 1,275).
(67) (Odysseus to Achilles, when the latter has shown too mucheagerness for fighting:)mê de houtôs, agathos per eôn, theoeikel'Akhilleunêstias otrune proti Ili huias AkhaiônTrôsi makhêsomenous."Most worshipful Achilles, the men have not breakfasted, andgallant as your are yourself, you must not order them to march onTroy and fling them like this at the enemy." (R. 19,155).
These two examples have in common with (65) that they are dissuasions.72 Theydiffer, however, from (65) in the felicity conditions of the dissuasive speech
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 203/318
PARTICIPLES Π : FORMULAS AND METRICS 193
act.73 In (66) and (67), the dissuaded activity is something which is exclusivelyattainable for men of high rank, and the addressees clearly belong to this rank.In (65), on the other hand, the addressee does belong to to the high rank, butthe activity in question is not particularly something for which one has to be
powerful. Everyone is able to cheat, regardless of power and esteem.
Now, this difference in speech act situation corresponds with a differencein the concession of 'participle + per'. In (66) and (67), the opposition Hes between the power of the addressee to perform the dissuaded act and the dis
suasion itself. As commander-in-chief, Agamemnon has, of all persons, the power to deprive Achilles of his girl. And his being commander-in-chief is intimately connected with his being agathos. Thus, we may say that qua agathos
he has the power to take away the girl. This is reflected in the request, in whichthe qualification agathos appears as a concessive modifier: 'do not, in spite ofyour being agathos, take away the girl.
Likewise, in (67) the dissuasion is addressed to someone who is in a position to do what is dissuaded. Being the best warrior of the Greeks, Achilles,too, is agathos par excellence, and he has a right to urge the Greeks on to fightunder any circumstances. Accordingly, in the request to Achilles not to withhold the soldiers their breakfast, the being agathos occurs as a concessive element. In (65), on the other hand, the opposition, which is the basis for aconcessive element, is not between the power of the addressee to perform thedissuaded act and the dissuasion itself, but between the nature of the dissuadedact and the moral and social stature of the addressee, which is something verydifferent.
Within the scalar-concessive framework of the previous chapter, agathos
pereôn as it is used in (66)-(67) is prototypical. It is an analytic participial phrasewith an adjective allowing of superlativeness (see 4.3.2.1); furthermore, in
being concessive with respect to the request as a whole, it is syntactically inde pendent and modally autonomous. To this characterization, we may now addthat agathos per eon as used in (66)-(67) is a perfectly integrated formula. Wemay even say that it is a formula in two senses of the word. First, agathos pereon conforms to a consistent and functional localization pattern. The localization of the phrase between positions 3 and 7 and the concomitant tri-partitestructure of the verse (see 5.2.1.2 and 5.2.2 sub Π . B1)) provide a sufficientrhythmical basis for the functioning of agathos per eôn as a formula.
But at the same time, agathos per eôn in (66)-(67) is a formula in a lesstechnical sense of the word. It is used as a stereotyped politeness formula whichtypically occurs in requests to people of high rank.76 Scalarle per' has its full force
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 204/318
194 LINGUISTICS AND FORMULAS IN HOMER
in this formula. It underlines the remarkable nature of the request: to dissuadean illustrious person from something which Hes in his power does not come asa matter of course. All this is very different in (65). The command here is nota polite request, but an angry reaction to Achilles' immediately precedingwords ("Admittedly, there can be no immediate compensation for the loss ofthe girl, but we will compensate you fourfold in the future"). This is an entirely different speech act, one in which poHteness formulas are out of place.
My conclusion is that agathos per eon in (65) is a less than perfectly in
tegrated formula. Being devised originaHy as a poHteness formula, it is lessfitting when there is no question of poHteness. The formular usefulness of thestereotyped introduction of requests directed to persons of high rank has beena sufficient condition for the use of agathos per eon in the context of (65). Butthe less than perfect integration does not imply that agathos per eon is meaningless or not in accordance with the context. The meaning of phrases is just asgradient an affair as the integration of the formulas they constitute. And theless than perfectly integrated status of agathosper eôn as a formula correspondswith a less than prototypical status of the phrase as an instance of the expression-type 'participle + per'.
I give one other example of semantic formular integration. In Odysseus'narrative to the Phaeaceans of his adventures (books 9-12 of the Odyssey), the
phrase kaka per paskhontes ('though suffering hardships') occurs six times. Itis three times complemented in metrical space by hetairoi ('comrades'). TheP2-formula per paskhontes hetairoi is typically used when Odysseus addresses his men, in combination with the P1-formula keklute meu muthôn ('lis-ten to my words'). The specific rhetorical function of the concessive participial
phrase in the address is to convey that what wffl be said is unpleasant for themen in their sorry state, but that it has nevertheless to be said. This is the casein the following example:
(68) (Odysseus and his men have just escaped the horrors of Scylla andCharybdis. They are now in sight of the island of Hyperion theSon-god. Odysseus then remembers the urgent advice of the seerTeiresias to avoid this island. So he has to dissuade his men fromdisembarking. He addresses them as follows:)keklute meu muthôn, kaka per paskhontes hetairoi"Friends, Hsten to my words, in spite of your hardships." (Od. 12,
271).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 205/318
PARTICIPLES II: FORMULAS AND METRICS 195The formula kaka per paskhontes may be characterized as fully integrated; itsconcessive meaning is appropriate, in view of the fact that what Odysseus hasto say is very unpleasant, for the men are very eager to disembark. The opposition which Ues at the basis of the use of per is thus between the sorry state ofthe addressees and the request itself.
Now the same line recurs somewhat later in the narrative (Od 12, 340).In the meantime the men, disregarding Odysseus' advice, have disembarked.They have promised not to touch the cattle of HeHos. However, a storm
prevents them from sailing further. When they are running out of supplies,Eurylochus, disregarding the warnings, urges his friends to slaughter some ofthe sacred cows, addressing his friends in the manner of (68). In this address,the kaka per paskhontes-formula is semantically less integrated, for what is saidis not something unpleasant, which has nevertheless to be said, in spite of themen being in a sorry state. Instead, the opposition Hes between the men's
present troubles and the content of the adhortation: Ί know that you are in badshape, but now listen to my words, for I have a solution.' This difference inspeech act situation seems to be connected with a difference in integration ofthe kaka per paskhontes hetairoi as a formula, and hence, with a difference in prototypicaHty of the phrase as an instance of 'participle + per.79
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 206/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 207/318
NOTES TO CHAPTER 5
1) Parry (1928a; 1928b). Henceforth I will refer to the collected work (Parry 1971).
2) For an overview of the development of Parry's thought, see Hoekstra (1965: 9-15) aswell as Visser (1987: 5 ff.).
3) In this definition, "a group of words" replaces the term "an expression" of an earlierversion (Parry 1971: 13 = 1928a: 15). This replacement has been approved of by some (cf.Hoekstra 1965:12,25 ("single words are not be called formulae")), but to my mind it is not necessarily an advantage. I see no reason why single words should not be called formulas. See further
below, 5.1.3.
4) See for instance ex. (6) in 5.1.3 below.5) This involves using formulas as one of the criteria of orality tests. Another criterion is
the relative absence of enjambment. For orality tests, see Lord (1960), Peabody (1975: 1-9).There has been much critique on the straightforward connection between formularity andorality. See for example Kirk (1966b: 174) and Russo (1976:38-39). The point of these authorsis that formularity and literacy are not mutually exclusive. When a text contains formulas, itsauthor evidently has used in some cases one and the same phrase for one and the same purpose,
but this need not imply that he is as an oral poet dependent on the recurrence.
6) Of course, this statement comes near to being a platitude. Nevertheless, it is remark-able that in spite of the unanimity on the functionality of formulas in oral poetry, this has had little or no impact on the definition of formulas. See further 5.1.1 below.
7) Parry is inconsistent in his use of the term 'idea'. Sometimes the term seems to apply tothe meaning of phrases/sentences, but often he uses it to refer to the phrases/sentences themselves.
8) Cf. Lord (1960: 65): "From the point of view of usefulness in composition, the formulameans its essential idea; that is to say, a noun-epithet formula has the essential idea of its noun."
9) All diachronic and dialectal divergences (see 1.5) from the contemporary linguisticnorm could be shown to be functional in that they yield metrically different doublets which greatly enhance the fluidity and smoothness of oral composition.
10) That is, the epithet is in function of, and conducive to, the formular system: "the poet
was guided in his choice by considerations of versification and in no way by the sense", Parry(1971:149). See also 1971: 305: "The fixed epithet in Homer is purely ornamental. It has beenused with its noun until it has become fused with it into what is no more than another metricalform of the name."
11) In general, we may say that the meaning of language can only be conceived of in termsof language. As such 'meaning' is a recursive affair. See Haas (1962) for a philosophical accountof this aspect of semantics (p. 212: "What an expression means cannot be found as a separateentity beside the expression.").
12) Almost needless to say, this distinction does not yield two discrete kinds of formula.Again, we have to allow for gradience in the continuum between the kind of formula that differs
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 208/318
198 LINGUISTICS AND FORMULAS IN HOMERmost from ordinary language as regards the form-content dichotomy, and the kind of formulathat differs least in this respect.
13) This is not to say that epithets are meaningless. Rather, we have to say that their meaning is subservient to metrical-formulaic purposes. See further 5.2.2 below.
14) 1971: 304: "It is the nature of an expression which makes of it a formula" (emphasisadded).
15) See on functions and categories Dik (1978:12-13). Parry uses '(being a) formula' as aone-place predicate: it ascribes an intrinsic, context-independent property to a given phrase. Inmy approach 'formula' is two-place; it tells something about the relation between a phrase and
the context in which it occurs. To my mind, the very reason why there is controversy over thedefinition of 'formula' is that 'formula' is a function, rather than a category. A definition of a linguistic function should not contain any reference to the expressions that may in principle perform that function.
16) Notice that this distinction should not be confused with the distinction between literary(poetical) and ordinary language. Homeric language of course differs sharply from ordinary language, but then so does the language of Pindar or Aeschylus. In treating Homeric language as
poetic language, we have not yet paid due attention to the specific element which distinguishesit from other poetic language, viz. the formula.
17) A notable exception is Kiparsky (1976). Kiparsky sees the oral formula as the analogonof the bound phrases ('idioms') in ordinary language. Bound phrases are arbitrary collocations
of lexemes, whose semantics is 'non-compositional', i.e. the meaning of the collocation cannot be explained on the basis of the meaning of the constitutive parts. Syntactically, bound phrasesare either free or fixed. Working on the basis of the distinction of Hainsworth (1968) between
flexible and fixed formulas, Kiparsky compares the flexible formula with the syntactically freecollocation and the fixed formula, which can appear only in one single form, with the syntactically fixed collocation. On this basis, Kiparsky contends that formular language differs quantitatively (more bound phrases) and not qualitatively from ordinary language (1976: 88). This is avery interesting approach, but it fails to explain the specific function of formulas in an oral-formulaic diction. Working in a generative framework, Kiparsky is characteristically more interested in 'rules' than in 'function'.
18) See also Visser (1987), where noun-epithet formulas (as well as other formulas and
formula-systems) are described as a complex consisting of a constant (the name/noun) and avariable (the epithet). The variable is the element of the formula by which it is adapted to themetrical exigencies at hand. Visser (1987) is an interesting study which came too late to my attention to be of any use in the preparation of this chapter. Visser's approach to the formula, inwhich 'formula' as a discrete notion (in the Parryan sense of 'ready-made building block') isrejected in favor of an approach in which a distinction is made between the semantic/metricalnucleus of the formula on the one hand and more redundant material filling the surroundingmetrical space on the other,is highly congenial with my treatment of 'participle + per' in 5.2.2
below.
19) This was already stated by Parry himself (1971:311-2, in a passage where he providesformulaic parallels for the phrases in the first lines of the Odyssey). See further 5.2.2 below.
20) Notice that Parry's definition ("a phrase which is regularly used under the same metri-cal conditions") has raised the question as to whether it is essential for a given 'formula' to al-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 209/318
NOTES TO
CHAPTER 5 199ways occur in one and the same place in the verse. Elsewhere (1971: 309-10), Parry recognizes
the fact that a formula may be placed in more than one place in the hexameter, though he con
siders it a rare phenomenon. In Ingalls (1979: 89), on the other hand, it is stated that we may
speak of one and the same formula in the case of any collocation of words which regularly recurs,
regardless of its metrical shape and its position in the hexameter. To my mind, the discussion
about whether formulas are metrically 'free' or 'flexible' is bound up with the structural-
categorial approach to the Homeric formula. When one considers it important to delimit 'the
formula' with regard to what is not formular, it is also important to know whether or not two
identical phrases which are localized in different positions in the verse are instances of one and
the same formula. But once we have adopted a functional approach to the formula, the ques
tion has to be reformulated: is it useful for the poet to have more than one localization for oneand the same phrase at his disposal? The multiple localization is formular only when its
functionality can be shown.
21) Of course, the repetition in both verses involves the recurrence of an adjective just
after the trochaic caesura. As the two lines are only 10 lines apart, there may be 'short-range
association' at work in the recurrence of this 'verse-pattern'.
22) In the formular analysis of Homeric passages, Parry, followed by Lord (1960), adopted
the practice to mark the verbatim repeated phrases with a solid line, and the repetitions involv
ing less linguistic sameness with a broken line (1971:301). But later on (1971: 313), just before
mentioning ex. (6) as well as the pair teukhe kunessin and dôken hetairôi, he contended that "one
could make no greater mistake than to limit the formulaic element to what is underlined."
23) See Russo (1963: 243-5). Many more phrase-patterns have been treated as two-vari
able formulas functioning as bucolic clausulas, for instance the type constituted by 'adjective +
noun': oinopa ponton, aglaa tekna, pioni demôi etc. See also Russo (1966: 239), Notopoulos
(1962:356-7).
24) Notice that oulomenên is a participle as to its form; as to its meaning, it is an adjec
tive.
25) In Hainsworth's often cited words (1964:157): "The vice of the term 'formula' to cover
structural features in the epic diction is that unless it is hedged about by more conditions than
are visible in the practice of present-day Homeric scholarship the statement that the epics are
nine-tenths formulae is likely to be vacuously, and so uselessly, true."
26) Minton (1965:245ff.) has analyzed the first 1000 lines of the Argonautica of ApolloniusRhodius (3th century ) in the same way as Russo did with the first 1000 lines of the Iliad (that
is, the same sample as O'Neill (1942) had used for his metrical quantifications in both poems),
and he claims that theArgonautica "reveals a distribution that is close to, and in some cases al
most identical with those indicated for Homer." Packard (1976) has probed the Posthomerica
of Quintus of Smyrna (Late Antiquity: 5th (?) century AD) for 'structural formulas' and con
cludes that here, too, there is no clear difference with the Homeric usage of phrase-patterns. To
this kind of objection Russo (1963:223-4; 1976:33-4) replied that the practice of literary poets
maybe seen as imitation of the early Greek hexameter. Kiparsky (1976:89), however, contends
that localization of word-category-types is a tendency of metrical verse in general.
27) See also 1966:227, where the choriambic participle placed at the beginning of the line
(see (9) in the text) is characterized as "in itself formulaic, because it plays such a frequent partin the poet's style."
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 210/318
200 LINGUISTICS AND FORMULAS IN HOMER28) Notice that Parry, too, did not consider mere repetition the essence of the Homeric
formula. But he used this fact to argue in a direction which totally differs from my argument
(1971:304): "It is the nature of an expression which makes of it a formula, whereas its use a second
time in Homer depends largely upon the hazard which led a poet, or a group of poets, to use it
more than once in two given poems of a limited length." (Emphasis added).
29) This is, of course, a formulation which is firmly embedded in the Parry-Lord way of
thought. It seems to me, however, that this kind of functionalism gradually receded to the back
ground in Parry's formular analysis of the opening lines of the Iliad and the Odyssey (1971:301
ff.). Here the concept of mere repetition has replaced usefulness and functionality. For instance,
the similarity of teukhe kunessi and dôken hetairôi seems to me wholly devoid of functional, and
hence, formulaic interest. Notice that Lord (1960:291) called teukhe kunessi 'nonformulaic', and
that Russo (1963: 245) had to admit that this structural formula (see (8) in the text) is not very
common.
30) Here we have, incidentally, another reason why Russo's structural formula is an un
satisfactory concept: it fails to differentiate the formula both from what is within and from what
is outside the diction.
31) Notice that' | ' in the scheme are alternatives. A verse cannot have at the same time a
penthemimeral and a trochaic main caesura.
32) This has the effect that in the bipartition of the verse neither the ratio 2:1 nor the
ratio 1:2 can occur.
33) Beekes' decision to deny position 3 and 7 metrical significance implies a rejection ofan influential way of thought in hexameter metrics, the 'four-colon theory'. Devised by Fraenkel
(1926), modified by Porter (1951), discussed by Kirk (1966) and advocated by Ingalls (1970,
1976), this theory assumes the existence of no less than three caesuras, which yields the concep
tion of a quadripartite hexameter. In Fraenkel's original theory, the first caesura falls at 1,1.5,
2 or 3, the second at 5 or 5.5 and the third at either 7 or 8. Beekes refutes this theory, showing
that the alleged first caesura has no structural/metrical significance, whereas the struc
tural/metrical significance of word end at 8 can and should be accounted for in other than
caesural terms. The four-colon theory amounts to a conflation of the actual words and phrases
filling the hexameter with the underlying rhythmical structure. Apart from this, the four-colon
theory is unnecessary, as Beekes (1972:4) states, in that O'Neill's data can easily be accounted
for without assuming the existence of caesuras other than or T.
34) Another of Hoekstra's examples is antitheos, which is both lexically and metrically
wholly equivalent to isotheos. Just as isotheos it does not conform to O'Neill's pattern: its
preferred position is 9 rather than 11. This is due to the formular system 'antitheos + proper
name' localized at verse end, for instance antitheos Thrasumêdês. An example comparable to
isotheosphôs is hômeteron do ("our house." For 'heav/ monosyllables placed at the end of the
line (violating rule 6) above), see Van Raalte (1986:88-9).
35) The apparent violation of the rule forbidding word and at 11 (see Beekes' rule 6) is
considerably mitigated when we recognize this.
36) The frequent divergence of words from O'Neill's figures for the word-type to which
the word belongs pomts to the fact that the linguistic (lexical) identity of words cannot possibly
be eliminated as a factor in the localization.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 211/318
NOTES TO CHAPTER 5 201
37) O'Neill's policy amounts to discarding the factor of gradience in the phonetic discrete
ness of words. Between the prototypically discrete word and the prototypically non-discrete
word a continuum may be construed.
38) According to Porter (1951:59) this happens, in Homer, in 1.2% of the cases. See also
Van Raalte (1986: 80-1).
39) Often proper names are involved here, for example: ê Aias | ê Idomeneus \ ê dios
Odysseus (Il 1,145).
40) Of course, localization in 7 is negligeable because in that case the word would bridge
the caesura).
41) These data can be easily gathered from Prendergast ([1875] 1962).
42) He may even deliberately change the word-type of single words. In doing this he creates
the artificial forms which are the trademark of Witte and Meister's notion of Kunstsprache. Thus,
as regards versification, we may say that what artificial morphology is on the level of the word,
redundant semantics is on the level of the formula.
43) This is not to say that the epithet is meaningless in Homer. But the meaningfulness or
appropriateness of a particular epithet in a given situation or context depends on the quality and
ingenuity of the poet in question; the appropriateness is not inherent in the formular system. A
good poet can make a highly significant use of names and epithets (see Edwards (1987:120-123)
on 'word-play and significant names' as well as Hoekstra (1981: 57 ff.)). To stress the poetic
quality of the epithets in Homer tends to become more and more popular in Homeric criticism
(e.g. Vivante (1982)), undoubtedly in reaction to the way the epithet was dealt with in Parryancriticism. But significance and 'redundant semantics' need not exclude each other. They are in
deed incomparable magnitudes.
44) For a similar approach to formulaic phenomena in Homer see Visser (1987).
45) Some case-endings constitute an exception to this rule: -oisi (plural dative) and -oio
(genitive singular, a variant of.-ou).
46) Notice that akhnumenos kêr is a violation of rule 6) in (11) above. But the word end
in pos. 11 is considerably mitigated when we conceive of akhnumenos kêr as a composite rhyth
mical word which plays a role in the formular system of 'participle + pef.
47) Akhnumenos kêr. Il. 7, 428, 431; 19, 57; 23,165, 284, 443; 24, 773. Od. 10,67; 12,153,
250, 270; 22, 188; 24, 420. Khôomenos kêr. . 1, 44; 9, 555; 23, 37. Od. 12, 376. Notice that akhnumenos kêr has a naturally long final syllable (word-type -⋃⋃--). So it has a 100%
localization in 12. Akhnumenos per, on the other hand, has a final syllable which is long only at
verse end. Accordingly, it has more than one localization possibility, see below.
48) See//. 8,125,317; 15,651; 17,459.
49) The argument/complement function is once performed by a verb: kai essumenon per
alukhsai ('though eager to escape'), Od. 4,416.
50) Morphologically, anagkêi is dative to anagkê ('necessity').
51) Consider, too, the post-caesural part of the line, where the characteristic expression
of 'force' and 'necessity' (see (18) above) is extended from the ⋃-- final slot to the range of
an entire half-line. The whole line is wholly and thoroughly characterized by redundant seman
tics.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 212/318
202 LINGUISTICS AND FORMULAS IN HOMER
52) The metrical fluctuation between and Τ is often exploited for the declension of
precaesural formulas ending on a consonant-stem noun or participle. Cf. the pattern
cassoni aïssontos/-ti etc. See Hoekstra (1981: 46). The distribution of per in (20)-(21) is a
fine example of the highly desirable situation for an epic poet to be able to use a P1 /P2-formula
as a T1 /T2-formula or vice versa. See Hoekstra (1965) for many examples of the permutation of
P2 and T2 formulas.
53) Notice that the second half of the line in Od. 7,218 ( = ex. (21)) adds nothing new to
the content of the sentence as a whole. The P2-part of the line is one of the many 'sorrow-
formulas' which are often used to the point of semantic vacuoussness.
54) The unique damassamenoi per, which was discussed in 4.3.2.3 above belongs to this
localization, too. We may now say that apart from its semantic strangeness, it is also unusual
from the point of view of formular versification in that in the remaining open space at the end
of the verse the syntax of the sentence is continued.
55) Compare with (25): Od. 15, 214: mala gar kekholôsetai empês. Here the participle is
replaced by a finite verb, and pef by a propositional particle (gar); the remaining open metrical
space at the end of the line is occupied by empês (see below, B.l)b). As regards meaning, there
is of course an essential difference between this finite verbal form and the scalar-concessive par
ticiple in (25). As regards form and localization, however, both phrases conform to a similar pat
tern.
56) According to O'Neill's table 16 (1942:145) (i.e. the basic word-type of the
participle) has a 100% localization in 12.57) For instance, an infinitival argument of the participle: mala per memaôta makhesthai
("however much he was eager to fight", . 13,317), or a semantically redundant infinitive: mala
per khateousi helesthai ("however much we needed it (to take)", Od. 13,280). The unique phrase
mala per nostoio khatizôn ("however much he is longing to return", Od. 11,350) is a reversal in
this type, the participle filling the final slot and the argument occupying the place of the par
ticiple.
58) This kind of phenomenon was already described by Witte (1912). Often the fluctua
tion in Homer between forms of one and the same verb belonging to different tense/aspect stems
as well as between middle and active forms of one and the same verb is entirely devoid of seman
tic implications; it is subordinate to formula-flection. In the present case, meneainôn is simply
the nominative form of the memaôta-paradigm. But notice that here, too, the principles of thelanguage are not violated: in the case of stative/durative verbs the difference between the present
participle and the perfect participle is minimal anyway, because the oldest Greek perfect denotes
states rather than (past) actions. In being stative, the participles meneainônlmemaôta yield
prototypical instances of 'participle + per' (see the discussions of time-stability in 4.3.2).
59) The difference between mala and empês in participial phrases is often neutralized in
the way differences between peripheral instances of different categories are neutralized (see
1.4). Sometimes, however, there is a difference, for example:
nêos eüsselmous halad' helkemen, ophr' eti mallon
Trôsi men eukta genêtai, epikrateousi per empês.
"To drag our ships into the sea, and put the Trojans, who have beaten us anyway, in an
even better position." (7/. 14,98).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 213/318
NOTES TO CHAPTER 5 203
a translation in line with the scalar-concessive mala per ('however much') is less
suitable. This seems to be due to the presence of matton ('still more'). Empês seems to be
used here in accordance with its etymology ('in all cases', 'in any case'). The example properly
belongs to the word-type (see ex. (32)).
60) Notice Parry's (1928a) insistence on the economy of formular systems. In the noun-
epithet formulas, 'doublets' are rare.
61) As well as, occasionally, elided forms: per eont'.
62) Notice that esti of which eôn is the participle is an enclitic. See 5.2.1.2 above.
63) It is interesting to note that O'Neills table for the word-type . to which eôn belongs
(1946:140, table 4) reveals a 12.9% localization in 5 (in the Iliad. Odyssey-. 10.1%) and a localization of no more than 1.5% in 9 (in the Iliad. Odyssey: 0%). These extremely low figures are due
of course to the fact that word end at 4.5 and 7.5 is consistently avoided. Thus we have here
another example of the divergence from O'Neill's figures which can be explained when we con
ceive of the word-type of a single word as subordinate to the word-group of which it is a part
(see 5.2.1.2 above). The localization of eôn in 5 and 9 in (33) and (34) is subordinate to the
localization of the whole word-type in 5 and 9.
64) For ligus per eôn agorêtês see ex. (42) in 4.4.1 above.
65) Two exceptions (out of 15 cases): Il 16,624; 24,609.
66) Only once does kai not precede: Il. 24,53.
67) Of course, the idea that 'irregularity' (of whatever kind) may be due to formular language use is in itself not new. Apart from the work of Parry, an early statement is Chantraine
(1953: preface): "L' Iliade et 1'Odyssée sont (..) des oeuvres littéraires qui se sont développés
par les procédés d'un style oral, au moyen de la technique des formules épiques. Il en résulte
que telle formule, employée en un passage donné, peut être transférée ailleurs où elle convient
moins bien."
68) Modification and flexion of formulas: 1971:197-201; juxtaposition: 1971: 202-221. In
the case of flexion of formulas, Parry could make use of the observations and explanations of
earlier scholars. Witte (1913:2223), for example, already noted that the brevis in longo in meropes.
anthrôpoi is due to the flexion of meropôn anthrôpôn. After Parry, the subject of modification of
formulas was extensively explored, in a diachronic perspective, by Hoekstra (1965).
69) See Bakker (1988b) for a discussion of hiatus after diphthongs and long vowels inHomer in terms of the flexion and juxtaposition of formulas.
70) Qua functionality, this pattern is equivalent to Russo's structural formula (see 5.1.3)
consisting of a choriambic participle placed at the beginning of the line.
71) Compare Od. 10,246; 14,142.
72) Furthermore, in the three examples, we have a 'negative present command' (see 4.3.4):
the addressee is requested either to stop doing what he is already doing (in (65) and (67)), or to
refrain from doing what he is about to do (in (66)).
73) Notice that agathos per eôn exclusively occurs in interactive speech and not in narra
tive passages. This is in line with the narrative technique of Homer: to call somebody agathos in
volves a moral, objective judgment about that person, and this is precisely something from whichthe Homeric narrator refrains. See further 3.2.4 above.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 214/318
204 LINGUISTICS AND FORMULAS IN HOMER
74) Cf. Adkins (1960:37). For another view see Sicking (1972:448). Here agathos per eônin (66) is interpreted as non-concessive.
75) Compare ex. (15) in 4.3.1 above.76) This characterization of agathos per eôn occurs already in Fraenkel's (1925) discus
sion of per. However, Fraenkel does not perceive the essential difference between (66)-(67) and(65). As another example of 'participle + per' as a politeness formula, Fraenkel mentions pinutê
per eousa in Od. 20,131; 21,103.
77) Or, in view of the repetition in ex. (67), we should perhaps say 'a less than fully integrated formular line, which makes no essential difference. The concept of semantic integra
tion of formulas seems to be the solution of what Denniston (1951:486), speaking about agathos per eôn, calls the "awkwardness of giving different interpretations to a word in identical lines indifferent contexts."
78) And three times by hikoisthe ('you may reach'), in a line meaning 'You may reachIthaca, though not without suffering hardships', see Od. 11,104, 111; 12,138.
79) The third occurrence of kaka per paskhontes hetairoi is Od. 10,189. This formula is,like (68), perfectly integrated, for Odysseus' words are unpleasant. It is important to notice thatin both cases the first line after Odysseus' speech is hôs ephamên, toisin de kateklasthê philonêtor ("So I spoke, and their heart broke"). The first line after Eurylochus' speech, on the otherhand, is hôs ephat'Eurulokhos, epi d' êineon alloi hetairoi ('Those were Eurylochus' words, andthe other men approved of them'). Thus the difference in function between the perfectly in
tegrated instances of kaka per paskhontes hetairoi and the less than perfect integrated instanceseems to be reflected in the reaction of the addressees.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 215/318
6 CONCESSIVE CONDITIONALS
6.0 Introduction
In the present chapter I discuss the use of per and kai in concessive condi-tionals. The subject of conditionals in connection with scalarity and even wasalready touched on earlier in this study, see 2.3.1.5-6 and 3.4.3. When usedtogether with conditionals, the scalar particle even can be used in two differentways. It may occur within the sentence that functions as the conditional. Thissentence then functions as the scope sentence of the particle. (We recall that
the scope sentence of a focus particle is the sentence which has a variable inthe place of the focus constituent of the particle, see 2.3.1.3.) But even mayalso be part of the main clause. In that case we have to say that the entire conditional subclause is the (sentential) focus constituent of the particle. In thesecond case we speak of 'concessive conditionals', The concessive conditionalin general and its occurrence in Homer in particular is the subject of the presentchapter.
The concessive conditional may be expressed in Homer in various ways: by placing per after the conditional subordinator (eiper) as well as by kai, whichis placed either before the subordinator (kai ei) or after (ei kai). Thus it ap
pears that once more kai intrudes in the discussion of per. There are obviously demarcation problems to be solved: is there any difference between ei perand kai ei, and if so, of what nature is that difference?
The study of ei per and kai ei once more shows that the relation between per and kai in Homeric Greek is complicated. After the discussion of the examples in which kai and per each have their own separate identity and function,kai functioning as the marker of extensional scalarity and per of intensionalscalarity (see 3.2.2 and 4.2.1), and of the examples in which the distribution ofkai in connection with per is determined by formular and rhythmical factors
(5.2.2), we now come to a third type of relation between the two particles. We
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 216/318
206 LINGUISTICS AND FORMULAS IN HOMER
shall see that there is a clear difference between the concessive conditionalswith per and those with kai, in that ei per and ei kai belong to different types ofconcessive conditional (see 6.2). Unexpectedly, in view of the picture of per andkai that has emerged so far, the ei kai conditional is nearer to superlativenessand scalarity than ei per. This surprising fact will have to be accounted for in adiachronic, rather than in a synchronic framework (see 6.3.3).
6.1 Concessives, conditionals and concessive conditionals
In chapter 4,I discussed the expression-type 'participle + per' as a concessive adverbial constituent (4.3). One of the most important characteristicsof 'participle + per' appeared to be the fact that it is a syntactically independentconstituent which is modally autonomous with respect to its main clause(nuclear predication). The resulting property which is relevant in the presentconnection is that 'participle + pef is factual irrespectively of the modality ofthe main clause (see 4.3.1).
This characteristic of 'participle + pef reflects an important property ofconcessive constituents in general. Concessives tend to create a factual environ
ment. In terms of modality (see 2.1.4), concessive adverbial constituents aretypically realis contexts. This is different in the case of concessive conditionals. The even if -conditional merely expresses, conforming to the meaning ofeven, that the state of affairs referred to by the main clause is something remark-able or unexpected (a 'superlative fact'). But to perform this function it neednot be factual. In other words, factuality is not a distinctive (prototypical)
property of the concessive conditional. A concessive conditional just may ormay not be factual. For example, in uttering
(1) Even if Achilles resumes fighting, we will win the war,
the speaker asserts that the side to which he belongs will win the war; Achilles'resuming fighting is preferably taken as one of a number of possibilities, ratherthan as a fact. It represents, owing to the scalar meaning of even, the least
favorable condition under which the war may be won. In
(2) Even though Achilles is resuming fighting, we will win the war
on the other hand, the concessive element has the same scalar meaning, in representing the most unfavorable possibility, but here it is factual. In the contextof uttering of (2), Achilles is supposed to be actually resuming fighting, and it
is asserted that despite this fact the war will be won.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 217/318
CONCESSIVE CONDITIONALS 207
Thus, concessive conditionals differ from concessives as regards factuality.Concessives are typically f
actual whereas concessive conditionals are not. Nowconcessive conditionals may be opposed to yet another expression-type, thesimple conditional. The basis for differentiation here is not the factuality ofthe adverbial element (subclause), but the factuality of the main clause. Thespecific difference between the concessive conditional and the simple conditional is that the main clause of the former is typically factual, albeit remark-
able/unexpected, as in (1)-(2), whereas the main clause of the latter is not.Someone uttering a statement in the format of 'If p, then q' does not normallyassert the actual truth of the main clause 'q' (as in the case of concessive conditionals), and he does not assert (or presuppose) the truth of the conditional'p', either.
Now, as is argued in König (1986), the three expression-types in questionhave to be seen as prototypical cases, rather than as distinct species. The semantic space between them is continuous, to the effect that the formal differencesmay be neutralized. As regards the relation between concessives and concessive conditionals, the neutralization lies in the fact that less prototypical concessive conditionals may be factual. And as for the neutralization of thedifference between simple conditionals and concessive conditionals, there existsome simple conditionals that allow of a concessive interpretation (see König1986: 238-39).8
In section 6.3, I discuss the Homeric concessive conditionals from the point of view of, among other things, their being factual or not, and in 6.4 wewill look at conditionals (mostly ei per) from the point of view of their beingconcessive or not. But first we have to introduce an important piece of typology in the field of concessive conditionals.
62 Kinds of concessive conditional
König (1986: 231) distinguishes three kinds of concessive conditional,which may be exemplified as follows:
(3) Even if Agamemnon apologizes, Achilles won't fight.
(4) However much you try to persuade Achilles, he won't fight.
(5) Whether you bring Achilles presents or not, he won't fight.
Ex. (3) exemplifies the kind of concessive conditional that functions as the sen
tential focus constituent of even in English. The proposition functioning as conditional ('Agamemnon apologizes') denotes a state of affairs which is highly
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 218/318
208 LINGUISTICS AND FORMULAS IN HOMER
unfavorable for the realization (or, as in (3), highly favorable for the non-realization) of the state of affairs denoted by the main clause. This is of courseexpressed by even, whose conventional scalar impHcature (see 2.3.1.1 above) isthat its focus constituent (sentential or not) represents the high point of a scale.This kind of concessive conditional will henceforth be called the 'even //-type'.
Ex. (4) exemplifies a different kind, which I will call the 'however much-type'. Here even (as it is used in English, that is) does not occur. The type exemplified by (4) has two inherent characteristics, superlativeness and
indeterminacy. Languages may differ as to which of these characteristics theyformally encode in their version of the type in question. English opts for indeterminacy: in the English version of the expression-type, the suffix -ever isused, which, when placed after a 'wh-word', has a universalizing function. Inthe German and Dutch versions of the expression-type, on the other hand, afocus particle is combined with a superlative expression following the 'wh-word' {wie sehr auch, hoe zeer ook). We shall see below that Homeric Greekexplicitly encodes the superlativeness, while Classical Greek opts for indeterminacy (behaving exactly like EngUsh).
The expression-type exemplified by (4) obviously has much in common
with the scalar-concessive participle discussed in 4.3. above. In both superlativeness plays a role and both are associated with a gradient, scope-independentscale. As 'participle + per' has to be discussed in terms of the semantics of intensional scalarity (see 2.1, 3.1 and 4.2), we may say that ex. (4) is connectedwith the scalar superlative, the starting-point of the discussion of scalarity inthe present study. The conditional in (3), on the other hand, has to be discussedin connection with extensional scalarity (see 2.2 and 3.2). The scale involvedhere is non-gradient and scope-independent, just as in the case of even with anextensional term as focus constituent (see 2.2.3). Thus, we may say that the two
expression-types exemplified by (3) and (4) are related to one another in thesame way as scalar superlatives are related to extensional scalar expressions{even + extensional focus constituent).
The concession in (5), finally, is based on a disjunction. The conditionalhere is related to yes/no-questions, in the same way as (4) is related to wh-ques-tions. In terms of morphology, there is a clear, language-independent connection between the expression-type exemplified by (5) and the conditionalsubordinator: the 'conditional disjunction' is often expressed by means of theconditional subordinator (Greek: eite...eite; Latin: sive...sive; Dutch: of...of).10
The semantic value of (5) is that the main clause is true irrespectively of the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 219/318
CONCESSIVE CONDITIONALS 209
truth-value (either t rue or false) of the proposition 'You bring Achilles presents'.
In the present chapter we are exclusively concerned with (3) and (4). Ex.(5) is irrelevant for the phenomena to be described. In the following section Iwill discuss per and kai in concessive conditionals. Both ei per (6.3.1) and kai ei(6.3.2) will be discussed with reference to the difference between (3) and (4)above as well as to the distinction between factual and non-factual. The discussion of ei per and kai ei in terms of the typology of (3) and (4) will yield the high
ly unexpected result that ei per -conditionals (when they are concessive, see 6.4)are exclusively of the type represented by (3), while the kai ei-conditional yieldsinstances belonging to (3) as well as instances belonging to (4). This is strange,for it amounts to a complete reversal of what we know of per by now: per has
been firmly established as the particle of intensional scalarity. Hence we do notexpect it to be consistently used in an expression-type that is connected withextensional scalarity, while the extensional scalar particle kai is used in an ex
pression-type that is connected with intensional scalarity. In 6.3.31 will suggesta diachronic explanation for this reversal.
6.3 Concessive conditionals in Homer
6.3.1 Per in concessive conditionals
The instances of ei per are either non-concessive or concessive conditionals. Non-concessive ei per will be discussed in 6.4; in the present section we areconcerned with the concessive instances of ei per. These are of the 'even if-type'exemplified by (3) above. As stated, this type can be characterized as aclause functioning as the focus constituent of even. The conditional may contain scalar terms, but this need not be the case.
In the following subsections I discuss various aspects of concessive eiper.The division into subsections does not correspond to any systematic division inthe material. Rather, I discuss instances which may, for some reason or other,
be set off against the other instances.
6.3.1.1 The discourse function of ei per
One of the salient properties of the expression-type 'participle + per' isthat it is never placed before the nuclear predication to which it is attached. Itis preferably placed behind. This feature should be seen as the consequence ofthe proper elucidating function of 'participle + per' : originally, 'participle +
per' must have functioned as an apposed constituent which explains the super-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 220/318
210 LINGUISTICS AND FORMULAS IN HOMER
lative intension of a preceding demonstrative (kai hôs, see 4.2.1). In the examples to be presented in the present subsection, which represent the mostcharacteristic and frequent use oieiper, on the other hand, the per-constituent(ei per-clause) is invariably placed before its main clause. This is the first difference between eiper and the other uses of per discussed earlier in this study.More will follow.
In the examples in question, the propositional connector gar ('for') has been added to eiper (ei per gar). It signals that the ei per-clause , together with
its main clause, serves as an argument for, or an elucidation or explanation of,the immediately preceding discourse. As such the discourse function of the ei per gar-instances is the same as that of any statement in which gar occurs.
(6) (Hector to Polydamas: 'You have no reason to fear the fighting, forthe following reason':)ei per gar t' alloi ge peri kteinômetha pantesnêusin ep'Argeiôn, soi d'ou deos est' apolesthai:ou gar toi kradiê menedêïos oude makhêmôn."Even if the rest of us are slaughtered wholesale by the Argive ships,
you need have no fear for your own safety - you are not the man tostand and fight it out." (Il. 12,245)
(7) (Agamemnon to Menelaus: 'The Trojans broke the truce. But theywill pay for if:)ei per gar te kai autik' Olumpios ouk etelessen,ek te kai opse telei, sun te megalôi apeteisan,sun sphêisi kephalêm gunaUcsi te kai tekeessi."For even if the Olympian postpones the penalty, he exacts it in theend, and the transgressors pay a heavy price, they pay with their Uves,
and with their women and their children too." ( . 4,160).
(8) (Thetis to Achilles: 'You need not worry about the body ofPatroclus':)ên11 per gar keitai ge telesphoron eh eniauton,aei tôid' estai khrôs empedos, ê kai areiôn."He could lie there through all the seasons of a year and still his fleshwould be preserved; indeed it might be fresher than now [Even if heUes there..]." ( Il 19,32)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 221/318
CONCESSIVE CONDITIONALS 211
Often the main clause opens with the adversative particle alla. This seemsto be a typical rendering of the opposition which is present in any concessiveconditional:
(9) (The seer Calchas: Ί am afraid to offend Agamemnon':)
kreissôn gar basileus hote khôsetai andri kherêi:ei per gar te kholon ge kai autêmar katapepsêi,alla te kai metopisthen ekhei koton, ophra telessêi,en stêthessin heoisi"For a commoner is no match for a king whom he offends: even if
the king swallows his anger for the moment, he will nurse hisgrievance till the day when he can settle the account." ( Il. 1, 81).
(10) (Odysseus to Achilles: 'A man cannot fight without breakfast':)ei per gar thumôi ge menoinaâi polemizeinalla te lathrêi guia barunetai, êde kikhaneidipsa te kai limos."His heart may be set on fighting ([for even if his heart..]), but
exhaustion takes him unawares, he is attacked by thirst and hunger,and his legs give under him." ( Il. 19,164).
(11) (A simile: a leopardess faces a huntsman without fear:)ei per gar phthamenos min ê outasêi êe balêisin,alla te kai perì dourì peparmenê, ouk apolêgei
alkes, prin g' ê ksumblêmenai ê damênai."Even if the man gets in first with a cast or lunge and she is pierced
by a javelin, her courage does not fail her and she grapples with himor dies in the attempt." ( Il. 21,576).13
Like any other normal conditional which precedes its main clause, the ei per r-clause in (6)-(11) provides the 'domain' within which the main clause is asserted and interpreted. The proposition functioning as conditional is not asserted in its own right; it sets the scene for the main clause, so to speak.14 Thespecific function of per is to indicate that this 'scene' is a highly unfavorable onefor the main clause to make sense in. Normally, it does not make sense for themain clause to be uttered under the conditions specified by the conditional.But this is of course the very point of ei per as a concessive conditional.
The main clause of (6)-( 11) is not an assertion in its own right which makes
an independent contribution to the discourse. Rather, it serves, together withits conditional, as an extra argument for, or an elucidation of, some previously
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 222/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 223/318
CONCESSIVE CONDITIONALS 213
The particle de is commonly described as a co-ordinating particle whichmarks either a continuative or an adversative relation between two sentences.As the equivalent of and or but in English it is opposed to gar in the same wayas these particles may be opposed to for.18
Note that unlike gar in (6)-(11), de does not occur within the conditional;instead, it is placed after a term which is extracted from the conditional (ton de,toioutoi de). A partial explanation of this practice can be given in terms of themeaning of concessive conditionals. The relation between an even if -condition-
al and the immediately preceding discourse cannot without more ado becharacterized in terms of simple continuation. The even ¿/-conditional is typically directed to what immediately follows (viz. its main clause), not to whatimmediately precedes. The reason of this is that the main clause of a concessive conditional is typically uttered as a superlative fact (see 2.3.1.2), and notas the simple continuation of what precedes in the discourse. This is why ei perde does not occur and why it is difficult to find an appropriate discourse-context for and even if.19
When there is a continuative relation with the preceding discourse, as in(12) and (13), this is not a relation between the main clause as a whole and the
preceding discourse. Rather, it runs via apart of the sentence which functionsas conditional, viz. its topic. It is in combination with this constituent that designals the relation with the preceding discourse.
6.3.1.2 Eiper and a fortiori argumentation
In the present subsection I present an example in which ei per occurs in adissuasion. From the fact that a concessive conditional denotes the leastfavorable circumstance under which the state of affairs referred to by its mainclause may obtain, it follows that the truth of the proposition functioning as
concessive conditional ('p') normally implies the falsity of the main clause ('q').In less logical and more functional terms, concessive conditionals owe their effect to the fact that when it is a reasonable and useful thing to say that 'p' is thecase, it is normally not reasonable and useful to say that 'q' is the case.21 Giventhis concessive relation between 'p' and 'q', when the truth of 'p' implies (iscompatible with) the falsity of 'q', it follows that this applies even more to the
falsity of p. For the falsity of ρ is may be seen as the alternative lower value onthe scale associated with even if.
Now on account of these semantics, the even if -conditional of the type rep
resented by (3) above has clear rhetorical possibilities: it has an illocutionary potential which makes it suited for the performance of directive speech acts, that
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 224/318
214 LINGUISTICS AND FORMULAS IN HOMER
is, speech acts that are meant to influence the behavior of an addressee. Thisis of course not surprising: even alone has the same illocutionary potential, aswas shown in 2.3.1.2 and 3.2.2 above. And even if has to be described decomposi-tionally, viz. as the intersection of the meaning of even and //separately.
The account of even if in terms of directive illocutionary force seems to be the right approach to the following example:
(14) (Polydamas, a Trojan with mantic qualities, tries to dissuade hisfellowmen from attacking the Greek wall:)hôs hêmeis, ei per te pulas kai teikhos Akhaiônrhêksometha stheneï megalôi, eïksôsi d'Akhaioi,ou kosmôi para nauphin eleusometh' auta keleutha:
pollous gar Trôôn kataleipsomen, hous ken Akhaioikhalkôi dêiôsôsin amunomenoi peri nêôn."Even if by a great effort we succeed in breaking down the Achaeangate and wall, and the enemy give way, our retirement from the shipswill prove disastrous. The Achaeans will be fighting for their ships.They will kill numbers of our men." ( Il. 12,223).22
This is a piece of a fortiori argumentation: 'we will suffer considerable losses(even) when we succeed in breaking down the gate. It follows that this is themore so when (which is more likely) we do not succeed in breaking down thegate. So let us forget the whole thing'. This a fortiori rhetorical mechanism isentirely in line with the scalarity of even. But in the case of conditionals, wherethe focus constituent of even is sentential constituent, there is an important additional property. When a concessive conditional has a discourse function as in(14), the practical consequence is that it denotes a state of affairs which is
presented as unlikely to occur, or, in other words, towards the fulfilment of
which the speaker has a sceptic attitude.In normal concessive conditionals that are not used to perform a directive
speech act, the speaker is normally neutral as to the fulfilment of the state ofaffairs denoted by the conditional. That state of affairs is merely presented asthe most unfavorable circumstance for the state of affairs denoted by the mainclause to occur. The point of such a conditional is to add salience to the assertion of the main clause.
In concessive conditionals that are used for directive purposes, on theother hand, the conventional scalar implicature of even creates what may be
called an epistemic modality in the conditional: '(Even) if ρ (which I do not ex pect), then q. So a fortiori if not-p, then q'. Notice, incidentally, that the scep-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 225/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 226/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 227/318
CONCESSIVE CONDITIONALS 217
"(Thamyris) had boasted that he would win in a singing-match even
if the Muses themselves, the daughters of Aegis bearing Zeus, would
be his competitors." ( Il. 2, 597).
Unlike the instances of ei per discussed earlier in this section, this example hassome characteristics which are in line with expression-types of per that have
been discussed in earUer chapters. The conditional has scalar properties by it-self: Mousai may function as the high point of a scale (especially a scale whichis associated to a scope sentence in which singing is involved), and autos
('himself) is potentially scalar. Accordingly, the conditional as a whole mayin principle be conceived of as some pragmatic superlative (see 2.1.3): Ί willwin under the most unfavorable circumstances' being equivalent to Ί could beatthe strongest competitors'. The typical irrealis environment, in which specifictemporal reference is lacking and on account of which we may speak in termsof scalar superlatives at all (see 2.1 and 3.1.1), is formed by nikêsemen, a futureinfinitive.
In spite of this difference, (18) remains a genuine instance of ei per in thatit is an instance of the 'even if -type' of the concessive conditional represented
by (3) above. But there is something more to be said about (18). The exampleis odd from the point of view of grammar, because the modal particle an occurs in the conditional.. An (or its Aeolic counterpart ke(n)) is the norm in conditionals in which the subjunctive mood occurs. But an in an optativeconditional occurs only once in Homer, in Il. 2, 597 ( = ex. (18)).
Admittedly, in itself an in optative conditionals is not impossible or oddin Greek. But (18) is not a genuine example of this usage. Instead of describing the conditional clause purelyy in terms of the semantics ofan, I want to
propose a different solution. The oddity of an in (18) seems to be due to thefact that the phrase ei peran autai in (18), a bucoUc clausula, constitutes a caseof less than perfect (not to say bad) semantic formular integration (see for theconcept 5.3.2 above). The phrase appears to be a formula which occurs elsewhere. In each occurrence, the conditional has subjunctive, rather than optative mood marking, and an is semantically perfectly in place. Here is anexample:
(19) (A simile: a lion finds a stag:)- ⋃⋃ - mala gar te katesthiei ei per an auton
seuôntai takhees te kanes thaleroi t'aizêoi
"..and devours it greedily in spite of all the efforts of the sturdyhuntsmen and the nimble hounds to drive him off." ( Il. 3,25).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 228/318
218 LINGUISTICS AND FORMULAS IN HOMER
In the third example of the bucoUc clausula in question, autos is replaced by a form which has a different meaning, but a very similar form and sound:
(20) (Aeneas: 'My horses are very swift':)tô kai nôi polinde saôseton, ei per an aute Zeus epi Tudeidêi Diomêdeï kudos oreksêi
"We can rely on them to get us safely into Troy, even if Zeus givesDiomedes son of Tydeus yet another victory." ( Il. 5,224)29
Of course, the meaning of aute ('again') is different from that oí autos. Butfrom the point of view of formulas and versification ei per an aute is similar toei per an autai/-on in (18)-(19).
Thus it appears that the linguistically irregular presence of an in the conditional in (18) can be given an account in formular terms. The rhythmically
pleasing phrase ei per an autoslaute is used, as a formula, under various semantic circumstances.31 It should be noted that the less than perfect integration ofei per an autai in (18) has more serious semantic consequences than the lessthan perfect integration of the formula agathos per eon (see 5.3.2). The latterhas its paraüels, and it may be perfectly described in purely linguistic terms (see4.3.4), whereas eiper an autai in (18) involves a linguistically irregular and un
paralleled use of an. Notice that there is an important difference between (18)-(20) and the in
stances of ei per discussed earlier in this section: in (18)-(20) the conditional is placed after its main clause, while in the examples discussed earlier it is placed before. This order is of course in line with the normal use of 'participle + per',which as an optional constituent is preferably placed after its main clause. Accordingly, the concessive conditionals in (18)-(20) seem to have the same function as 'participle + per': they function as scalar-concessive constituents which
contribute to the salience of their main clauses, by signalling that the latter istrue in spite of something highly unfavorable. The difference with 'participle+ per 9 Hes, of course, in their non-factuality.
The instances of ei per discussed earlier in the present section are clearlydifferent. Here the sentence-initial eiper-clause is much more like a real conditional which functions as the topic ('theme') about (within) which the mainclause asserts something. In (18)-(20), on the other hand, nothing is assertedin the main clause about the conditional. Rather, it is the other way around, theconditional asserting something about the main clause, in the typical manner
of scalar-concessive constituents.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 229/318
CONCESSIVE CONDITIONALS 219
6.3.2 Kai in concessive conditionals
Despite a number of differences, in discourse function and in metricallocalization, the instances of ei per discussed in the previous section form ahomogeneous group in that they all belong to the 'even if-type' of concessiveconditional. Now, since I endorse a decompositional analysis of even if, in termsof even and if separately, it is interesting to study the use of kai, the normal andunequivocal Greek expression of even, in concessive conditionals. In the
present subsection I discuss kai ei (ei kai) within the framework developed so
far for the concessive conditional.The instances of kai ei in Homer may be divided in two ways. First, unlike
ei per, the kai ei-conditional is not exclusively used for the expression of concessive conditionals of the 'even if-type'; there are also instances of kai ei thatare of the 'however much-type'. Second, there are factual as well as non-factualinstances of kai ei. It will appear that kai ei is more often factual than eiper. Inthe following subsection (6.3.2.1) I discuss kai ei-instances of the even if-type,and in 6.3.2.2 instances of the however much-type. The distinction between fac-tual and non-factual will be dealt with in passing.33
6.3.2.1 Kai ei: 'even if'
Just as in the case of ei per, the kai ei-conditional normally does not contain elements or terms that are in themselves scalar as kai takes care of thescalarity and, hence, concession. But just as in the case of ei per, exceptions are
possible. Consider:
(21) (Athena to Odysseus:)kerdaleos k' eie kai epiklopos, hos separelthoien pantessi doloisi, kai ei theos antiaseie.
"What a cunning knave it would take to beat you at your tricks. Evena god would be hard to put it." (Od. 13,292).
(22) (Diomedes to Paris: 'You are a weakling and your shot did me noharm':)e t' alios hup'emeio, kai ei k' oligon per epaurêi,oksu belos pelettai, kai akêrion aipsa tithêsi."My weapons have a better edge. Even if they touch a man ever so
little, he is dead." ( Il. 11,391).
Ex. (21) is semantically compatible with (18) above. A term with scalar potential (theos, ' god') is used in a concessive conditional which is expUcitly marked
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 230/318
220 LINGUISTICS AND FORMULAS IN HOMERas such. In (22) negative polarity is involved. See also exx. (16) and (17) in 6.3.1.4above.
The following two instances do not contain scalar terms. But as regardsthe concession they are wholly compatible with (21) and (22):
(23) (Athena and Odysseus before the palace of Alcinous:)su d'esô kie mêde ti thumôitarbei: tharsaleos gar anêr en pâsin ameinônergoisin telethei, ei kai pothen allothen elthoi"Go straight in and have no qualms. For it is the bold man who everytime does best, even if he is abroad." {Od. 7, 52).
(24) (Odysseus (in disguise) to Telemachus: 'are you willinglyundergoing this outrageous conduct of the Suitors?':)ê ti kasignêtois epimempheai, hoisi per anêrmarnamenoìsi pepoithe, kai ei mega neikos oretai."Or is it your brothers who cannot be trusted to stand by as theyshould through thick and thin (litt, even if a great quarrel arises)."(Od. 16, 98).
The four examples above are instances of the non-factual concessive conditional of the 'even if -type'. In two cases, (23) and (24), the propositionfunctioning as conditional is non-factual anyway, irrespectively of its functioning as a conditional, because generic modality is involved (see 2.1.4 and 3.1.3above). Their main clauses are statements about the generic bold man and thegeneric brother, respectively; there is no specific temporal or spatial reference.
However, it is perfectly normal for kai ei to express a factual concessiveconditional with specific temporal reference. Of ei per there was only one such
an instance in the data-base; kai ei
yields more cases: out of 11 instances of kaieilei kai (of the 'even if -type') in Homer, 4 are factual. Consider:
(25) (Poseidon, in the shape of Calchas the seer to both Aiantes: 'It ishere that you have to make your stand, for here Hector is leadingthe attack:')to ke kai essumenon per erôêsait' apo neonôkuporôn, ei kai min Olumpios autos egeirei."You might fend him off from the gallant ships, for all his fury and
the encouragement he gets from the Olympian himself." (IÏ. 13,58).
(26) (Hector urges the Trojans not to fear Achilles too much:)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 231/318
CONCESSIVE CONDITIONALS 221
tou d' ego antios eimi, kai ei puri kheiras eoiken,ei puri kheiras eoike, menos d'aithôni sidêrôi."And I am going to meet him, though his hands are like fire, yes,though his hands are like fire and his heart like burnished steel." ( Il.20,371).35
Ex. (26) shows, together with the factual ei per-instance (15) above, thatwhen a concessive conditional of the 'even if -type' is repeated, as in (26), orwhen a second one is coordinated, as in (15), the second conditional can dowithout the focus particle. Ex. (25) is interesting in that a factual case of ei kaiis coordinated (by the sense) with an instance of participle + per' We may saythat Hector is essumenos ('eager', 'furious') because Zeus arouses him.
6.3.2.2 Kai ei: 'however much'
Kai ei differs from ei per in that it yields instances of the second type of theconcessive conditional as well, the 'however much-type', in which superlative-ness is involved (see (4) in 6.2 above). In the present subsection I present theexamples; in 6.3.3 below I will discuss the surprising situation that it is kai ei,and not ei per that is used for the expression of superlative concessive conditionals, whereas it is per, and not kai, that has a direct connection with super-lativeness and intensional scalarity.
The instances of kai ei which represent the 'however much-type' of theconcessive conditional contain without exception the particle mala (Very')which is followed by an adjective or an element with adjectival properties.36
The meaning of this particle is of course in line with the superlative characterof the type of concessive conditional in question: something is asserted to bethe case, in spite of the high degree to which some unfavorable circumstance
is the case. Some examples:(27) (Odysseus has thrown off his beggar's disguise, and aims an arrow
at the suitor Antinous. The latter is completely unprepared:)tis k' oioito met' andrasi daitumonessimounon enipleonessi, kai ei mala karteros eie,hoi teuksein thanaton te kakon kai kera melainan?"For who could guess, there in that festive company, that one man,however powerful he might be, would bring calamity and death tohim against such odds?" (Od. 22,13).37
(28) (Nausicaa to Odysseus: 'Go to the palace and acts as follows':)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 232/318
222 LINGUISTICS AND FORMULAS IN HOMER
ton parameipsamenos mêtros perì gounasi kheirasballein hêmeterês, hina nostimon êmar idêai
khairôn karpalimôs, ei kai mala têlothen essi"Slip past my father and clasp my mother's knees if you wish to makecertain of an early and happy return to your home, however far youmay have strayed." (Od. 6,312).38
These concessive conditionals (as well as their repetitions) are non-factual interms of the distinction made in 6.1 above. But there is a difference between
the two as regards the nature of the non-factuality. In (27) the non-factuality isdue to non-referentiality: mounon ('one man') does not refer to any particular
person; the point in (27) is the opposition between one man (whoever he is andhowever strong he is) and many feasting Suitors. In (28), on the other hand, theei kai/ -clause occurs in a stretch of interactive speech; its subject is the addressee. And, conforming to the typical non-factual character of the (concessive)conditional, the ei kai-clause does not say that the addressee (Odysseus) hasactually strayed far from home.
It is clear that, on account of the presence of superlativeness, as well as on
account of their position with regard to the main clause (postposed), concessive conditionals of the 'however much-type', like (27)-(28) above, have muchmore in common with scalar concessives of the type 'participle + per' than evenif -conditionals like the ei per -instances (6)-(15) above. In one case per actuallyoccurs in the ei kai -clause, confirming this semantic similarity:
(29) (Odysseus finds a shelter after his shipwreck:) phullôn gar eên khusis elitha pollê,hosson t' duo ê treis andras erusthaihôrêi kheimeriêi, ei kai mala per khalepainoi.
"There was an abundance of dead leaves, enough to providecovering for two or three men in the hardest winter weather." (Od.5, 485).
Per here forms a collocation with mala which is localized as in the case of participles of the word-type uu - - (see 5.2.2 sub I.B1)). Just as in the case of (27),the conditional occurs in a non-factual environment, where both spatial andtemporal reference is absent. Accordingly, this conditional has semantic affinity with a scalar superlative.
Just as kai ei-clauses of the 'even if-type', so kai ei-clauses of the 'however
much-type9
can be factual. Here are two examples:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 233/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 234/318
224 LINGUISTICS AND FORMULAS IN HOMER
The participial phrase in this example is semantically equivalent to non-factual concessive conditionals of the 'however much-type'. In terms of Oguse(1962), the participle lies in the modal sphere of the main clause (see also 4.1.2).
6.3.2.3 Kai ei or ei kai?
In the section on kai in concessive conditionals, I have treated kai ei andei kai indiscriminately. Strictly speaking, however, there must be a difference
between the two. Kai ei is the straightforward formal equivalent of even if (en
tire conditional as the focus constituent of kai), whereas ei kai seems to be properly used when the focus constituent of kai is one single word within theconditional. Something of this difference is clear when we compare ex. (25)to (26): in (25), a term in the conditional, Olympios autos ('the Olympianhimself, is a reasonable candidate for focus constituent of kai (see note 25above for 'self and scalarity), whereas in (26) the focus constituent is clearlythe conditional as a whole.
The investigation of the difference between kai ei and ei kai, however, isseriously hampered by the fact that the choice of either one in preference tothe other is in practice determined by other factors. Just as in the case of kai in
participial phrases, the order of ei and kai, and hence the question as to whatis the focus constituent of kai, is made subservient to metrics and versification.With two exceptions, all of the instances of concessive conditionals with kai
are placed just after the main caesura. Now when the main caesura is realizedas Τ (trochaic caesura, see 5.2.1.1 above), and the post-caesural formula is, accordingly, a T2-expression, we find kai ei (u-). And, conversely, when thecaesura is P(enthemimeres), we find ei kai ( - - ). The difference between thetwo expressions is subordinate to their function as the introduction of a T2- anda P2-phrase, respectively. Thus, the fluctuation between ei kai and kai ei ap
pears to be another adaptation of linguistic form to the exigencies of theverse. 42
6.3.3 Ei per and kai ei in a diachronic perspective
The problem which is the subject of the present section may be stated asfollows. There are two concessive conditionals in Homer, ei per and kai ei. Theformer is exclusively used for the expression of conditionals of the even if-type,
that is, concessive conditionals in which no superlativeness is involved, andwhose scale is, accordingly, scope-dependent and non-gradient (see 6.2). Kai
ei may express this type, too, but at the same time it may appear as a condition-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 235/318
CONCESSIVE CONDITIONALS 225
al of the 'however much-type' in which superlativeness is the salient characteristic.
This situation is strange, for one would expect the reverse situation, eiperexpressing the 'however much-type' and kai ei the 'even if-type'. This would bein accordance with the meaning and function of the two particles that hasemerged so far in this study. Per has close connections with superlativeness, inmany of its central uses, while kai is aparticle which is equivalent to the Englishscalar particle even, at least even as it is used in even if. The two particles may
be differentiated on the basis of the distinction between intension and extension (see 2.2.2), per being the particle of intensional scalarity and scalar superlativeness and kai the particle of extensional scalarity.
The solution to the problem posed by the unexpected distribution of ei per and kai ei over the two types of the concessive conditional has to be soughtnot so much in the meaning of per as in the change of that meaning. That is, ina diachronic framework. The relevant aspect of the diachrony of per, whichseems to be crucial for the assessment of the ei per -clauses in 6.3.1, may betermed 'disappearance of superlativeness and scalarity'.
We have met with this diachronic phenomenon before, in the case of the
loss of superlativeness in participial clauses (see 4.3.2). It could be argued thatthe use of 'participle + per' in Homer extends, in the diachronic dimension,
beyond the purely scalar. Many cases can be found in which 'participle + per'has lost its original superlativeness, to the effect that it has become a purelyconcessive modifier. This process is not confined to Ancient Greek. In manyunrelated languages, concessive elements are erstwhile scalar expressions.
Now, the concessive conditionals introduced by ei per discussed in 6.3.1have to be located, in the diachronic space, in the neighborhood of the non-su
perlative and non-scalar instances of 'participle + per'. They do not involve su
perlativeness and normally they do not contain terms with scalar potential.Consequently, when they are discussed in connection with scales and scalarity,we have to conceive of the alternative lower on the scale as the negated counter part of the sentence in question (see also 2.3.1.6 above). But this is preciselythe way to discuss non-superlative cases of 'participle + per 9 in scalar terms:varying on the theme of ex. (23) of ch. 4, we may say that the reason why
(33) Although he was on his own, he was not afraid
may be discussed in scalar terms is that the person in question would not beafraid a fortiorì if he were not on his own.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 236/318
226 LINGUISTICS AND FORMULAS IN HOMER
Thus there is a clear semantic similarity between non-superlative 'participle + per' and concessive conditionals introduced by eiper. The use of bothmust have become possible when the original scalar meaning of per was eitherfading or had already disappeared from the language. But there is also a difference between the two. Non-superlative 'participle + per' is an organic partof an expression-type which comprises scalar expressions as well; its scalar
provenance may be easily traced (see 4.2 and 4.3). Eiper, on the other hand,has no direct scalar parentage; it is a 'newcomer' in the diction. This seems to
be confirmed by the fact that eiper is common in Homeric similes.There are different types of simile in Homer,43 which can be ordered bydegree of increasing complexity. It may be confidently supposed that the increase of complexity has a diachronic orientation: the more complex a simileis, the more recent it is in the diachronic cross-section of the Homeric language.The simplest type may be exemplified by an expression like 'Achilles raged likea Hon'. A more elaborate type consists of an extension of this simple type, inwhich the relevant essential property (the so-called 'tertium comparationis')of the entity compared (often an animal) is furnished. The extension is veryoften in the form of a non-restrictive relative clause modified by the particle te
(see 3.1.3).44 An example is ex. (9) in3.1.3: "They were excellent speakers, likecicadas, which...".
In the most elaborate type of simile, the extension has evolved in an oftenminute image of the comparing entity (animal), which has become inde
pendent, both from the point view of content and from the point of view of syntax. Shipp ( 1972) has shown that the influx of 'late' forms is greatest in thesesimile extensions as well as in other digressive contexts. In other words, it ap
pears that the elaborate simile extensions are one of the most recent (context-types in Greek epic diction.
Now ei per frequently belongs to the independent part of an elaboratesimile. I think that it is not too hazardous to suppose that the regular occurrence of ei per in these innovative passages may be treated as an index of therelative 'lateness' of ei per. The lateness, in its turn, would seem to explain itsnon-superlative character.
The diachronic dimension is essential for the description of concessive ei per in Homer. Not only comes concessive ei per into being at a relatively latedate, it is also diachronically unstable, in that it has not had a long life46: ei perused as a concessive conditional does not occur after Homer.46 The only Greekconcessive conditional after Homer is kai ei.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 237/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 238/318
228 LINGUISTICS AND FORMULAS IN HOMER
tual, see (32) above. In the present section, I deal with the difference betweenconcessive conditionals and simple conditionals. This difference will be usedas a background for the discussion of the concessive vs. the non-concessive instances of ei per.
Just as the difference between concessives and concessive conditionalsmay be neutralized, so may the formal difference between concessive conditionals and simple conditionals. However, the neutralization as it is discussedin the literature (König 1986) works in the opposite direction: the concessive
conditional does not adopt a property of the opposed term (as in the case ofthe neutralization of the difference between concessives and concessive conditionals), but the opposed term adopts, conversely, a property of the concessive conditional. The formal difference between concessive and simpleconditionals may be neutralized in that simple conditionals that are not overtly marked for concession (by a focus particle, for instance) may allow, in certain contexts, of a concessive meaning.
König (1986: 238-39) lists three contextual conditions under which thedistinction may be neutralized. Neutralization occurs (i) when the conditionalcontains a scalar term and the conditional is not overtly marked for concession;
(ii) when the conditional occurs in a question, and (iii) when the main clauseis negated and contains an anaphoric pronoun which refers back to the conditional. Of each an example:
(34) I will not renounce my feud if you give me the most valuable presents.
(35) Will you stay confident if the Trojans are victorious?
(36) If Achilles refuses to fight, the Achaeans are not discouraged by it.49
However, the neutralization pattern displayed by these examples does notapply to ei per. In the present chapter we are not dealing with simple conditionals which come to have the function, in some particular discourse contexts, ofa concessive conditional. Ei per is not a simple conditional. Accordingly, to dealwith the non-concessive instances of ei per in a satisfactory way, we have to lookin what ways a conditional which is overtly marked for concession, like ei per,can lose its typical concessive meaning. I will argue that the non-concessive instances of ei per have to be discussed partly in synchronic and partly indiachronic terms. In a synchronic framework, there are semantic lines to befound along which ei per can lose its typical concessive meaning, adopting
another typical function instead. As for diachrony, we have already seen that
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 239/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 240/318
230 LINGUISTICS AND FORMULAS IN HOMER
right, namely when the diachronically unstable meaning of ei per evaporates.Furthermore, Prototype Theory comes into the picture once we realize that theinstances of ei per that have become 'sceptic' in their own right and in whichconcession is (near-)absent, may be viewed as one extreme of a semantic (lexical) continuum. The other extreme of the continuum is formed by the prototypically concessive instances. In between we may localize instances by degree ofdecreasing concession and/or increasing scepsis.
6.4.3 From concessive to non-concessive eiperConsider now, against the background of the previous section, the follow
ing example:
(37) (The seer Teiresias to Odysseus in the Netherworld: 'If you leavethe cattle of the Sun-God untouched, you may reach Ithaca:')ei de ke sinêai, tote toi tekmairom'olethronnei te kai hetarois: autos d' ei per ken aluksêis,
opse kakos neiai, olesas apo pantas hetairous."But if you hurt them, then I warrant that your ship and yourcompany will be destroyed, and if you yourself do manage to escape,
you will come home late, in evil plight, upon a foreign ship, with allyour comrades dead." (Od. 11,113 = 12,140).
The conditional here may in principle be read as a concessive one, but there ismore to it than that. There is a difference between ei per in (37) and the
prototypically concessive instances of ei per: in (37) the conditional does notexpress a proposition which is unfavorable for the truth of the main clause: between 'Odysseus manages to escape' and 'Odysseus comes home in evil plight
and without his comrades' there is no contrastive relationship; the former isnot unfavorable for the latter. Rather, it is a necessary condition for the utterance of the latter: it does not make sense to say that Odysseus comes homelate unless he has actually escaped. In presenting the 'escaping' in the form ofa conditional, the speaker effectively conveys detachment: 'it is far from certain whether you escape at all, but IF you do, you will reach home in a particularly wretched way'53. Thus what was called above the sceptic tone of eiperis more important here than the concession, and the example has to be treatedas a transitional case between the prototypical concessive ei per and the
prototypical 'sceptic' ei per.54
The following example is a transitional case too, an instance of the 'concessive-sceptic' conditional:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 241/318
CONCESSIVE CONDITIONALS 231
(38) (Idomeneus to Meriones: 'Don't be afraid that I consider you acoward. For if we were lying in ambush, you would belong to the
brave sort of man':))oude ken entha teon ge menos kai kheiras onoito.ei per gar ke bleio poneumenos êe tupeiês,ouk an en aukhen'.opisthe pesoi belos oud' eni nôtôialla ken e sternôn ê nêduos antiaseie."Nobody on such an occasion would think Ughtly of your daring and
your strength! Even if you were hit in the action by an arrow or spear,it would not be behind on your neck or back that the weapon wouldfall; it would strike you in the chest or belly." ( Il.13,288).55
Again, the ei per -clause does not contain, as a real concessive conditional does,something that is unfavorable for the main clause to be true: being hit by anarrow cannot, of course, be viewed as an unfavorable circumstance for not beinghit by an arrow in the back. Rather, the function of ei per, just as that of even ifin the translation, is to introduce a sceptic tone. The speaker rather reluctantly has to admit, as it were, that being wounded remains after all a possibility,
even for the brave man.In the following examples we are near to the 'sceptic' extreme of the con
tinuum. In (39) the conditional is 'sceptic' in the true sense of the word, and(40) can be perfectly accounted for in the 'concessive-sceptic' framework:
(39) (Odysseus is expressing outright scepsis to Athena as to the outcomeof the struggle with the Suitors: 'Being alone, how can I ever defeatthem? And, moreover:)ei per gar kteinaimi Dios te seihen te hekêti,
pêi ken hupekprophugoimi? ta se phrazesthai anôga."Suppose that by Zeus ' grace and yours I bring about their deaths,to what safe refuge can I fly? These are the problems I should likeyou to consider." (Od. 20,42).
(40) (From the Ares-Aphrodite story: Ares and Aphrodite are lyingcaught in Hephaestus' net. Hephaestus refuses to release them,fearing that Ares will not pay him the fine: 'Who will pay if Ares doesnot pay?' Poseidon then says:)
Hêphaist', ei per gar ken Arês khreios hupaluksasoikhêtai pheugôn, autos toi ego tade tisô."Hephaestus, said Poseidon the Earthshaker, if Ares does repudiatehis debt and abscond, I myself will pay you the fine." (Od. 8,355).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 242/318
232 LI NGUISTICS AND FORMULAS IN HOMERIn (39) the environment as a whole in which the conditional occurs is sceptic.In (40) the topic of conversation are the chances that Ares will pay Hephaestus the fine after he has been released. Poseidon's contribution to the conversation is: 'It is likely that Ares will pay the fine. But in case he does not, I will
pay the fine myself. In both (39) and (40), the connection with concession isweak.
Now it might be asked whether the continuum ranging from 'concessive'to 'sceptic', within which the instances oieiper can be localized, is diachronic
or synchronic in nature. In other words, is the fluctuation displayed by the instances of ei per in Homer due to a variability which is inherent in the meaningof ei per, or is it due to a change in that meaning? My answer would be that itis due to both. It seems that there is always some fluctuation inherent in themeaning of even if -conditionals, depending on how they are used in discourse.
Even if in English, too, can be used when the conditional in question is not a prototypical concessive one. See for example the translation of (38) above. Buta diachronic aspect is involved as well. This appears from the fact that the oneextreme of the continuum, the (prototypical) concessive ei per, simply ceasesto exist at some time, while 'sceptic' eiper continues to be used after Homer.Thus, beside a semantic continuum which allows fluctuation between concessive and sceptic use of the conditional, there is also a diachronic shift from concessive to sceptic, to the effect that the former becomes extinct at some time.
The examples discussed in the present chapter do not cover all the extantinstances of ei per in Homer. There are examples that are neither concessivenor sceptic and that cannot, accordingly, be localized in the semantic continuum ranging from concession to scepsis. However, those examples do not
properly belong to the present chapter. They should be dealt with within thewider framework of the meaning of per in general, and the periphery of the
semantic field of per in particular. As such they are best discussed in the nextchapter, along with the other instances of per which do not conform to anyrecognizable semantic or discourse-pragmatic pattern.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 243/318
NOTES TO CHAPTER 6
1) In 2.3.1.5 we saw that the scope sentence of even may also comprise the clause functioning as conditional plus the main clause.
2) Of these two possibilities, kai ei is best in accordance with the use of kai as a focus particle, because, like even, kai normally precedes its focus constituent. Thus, as to its form kai ei ismore in accordance with even if than ei kai. It may be asked whether there is a difference inmeaning between kai ei and ei kai. This question will be dealt with in 6.3.2.3 below.
3) Of course, in strict linguistic terms and apart from the metrical-formular framework,the function of kai in participial phrases with per has to be stated in terms of strengthening.
4) In other, more logic-based terminology, it is sometimes said that concessives are im plicative. In König (1986:231), for example, it is stated that a concessive expression of the formthough p, q "entails" the actual truth of both the "antecedent" (viz. the concessive adverbial constituent) and the "consequent" (viz. the main clause).
5) For the oppositions between concessive conditionals on the one hand and concessivesand conditionals, respectively, on the other, see König (1986).
6) Note that the main clauses of (1)-(2) are called factual while they contain difuture predicate, a tense that was discussed in 2.1.4 as a typical irrealis phenomenon. But the factuality of
futures depends ultimately on the nature of the speech act in which they occur. Confident predictions like the main clause in (1)-(2) are (meant to be) no less factual than utterances in whichunequivocal realis tenses/moods occur. See also note 17 of chapter 2.
7) In 1.4I argued that normally in linguistics, categories are discrete as to their form butcontinuous as to their meaning, to the effect that we should not speak of 'fuzzy edges' and overlap between categories but of the neutralization of the formal differences between them.However, in the case of concessives and concessive conditionals there may also be formal overlap: the (factual) concessive may be expressed in the form of a concessive conditional ( = even+ the conditional subordinator). Examples are et-si in Latin and even though in English (seeKönig 1986: 240). Very often diachronic factors are involved in this shift of concessive conditionals to concessives {though in English is originally non-factual).
8) A different kind of neutralization between conditionals and concessive conditionals isthe existence of conditionals whose main clause is factual, just as the main clause of a concessive conditional.
9) Ever as it is used in (4) is sometimes called, like universal any, a 'free-choice quantifier',in that it indicates a free choice from any number of possibilities. See Quirk et al. (1972: 751)and König (1986: 231). The suffix -ever is distributionally not confined to concessive sentences,where a contrast is expressed or implied. It may also occur in 'indifferent' relative clauses ('I'lldo whatever you like'). See further 6.3.3 below.
10) Dutch of (cf. German ob) is etymologically connected with English if. In a number oflanguages, conditional subordinators are etymologically derived from and/or semantically connected with interrogatives. See Haiman (1978: 565), Traugott (1985: 291). And often, the indirect interrogative particle is cognate to the disjunctive particle.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 244/318
234 LINGUISTICS AND FORMULAS IN HOMER11) en is the contracted form of ean ( =ei 4- an). As such it is relatively recent, diachroni-
cally, and it could be a clue as to the diachronic status of ei per (see further 6.3.3 below).
12) This 'apodotic' use of alla poses some problems. One might suppose that alla intro
ducing an apodosis points to co-ordination (see Ruijgh 1971:712,785). And the adversative co-
ordinate relation between two (independent) clauses is indeed semantically compatible with the
concessive conditional. However, many ei per-clauses contain the subjunctive mood, the subor
dination-mood par excellence. This would seem to point to an adverbial status of alla
('nevertheless'). Perhaps the dilemma may be solved in diachronic terms: synchronically,
apodotic alia is adverbial, whereas diachronically it has the function of a coordinator.
13) The remaining instances of concessive ei per gar. Il. 4,55; 8,153; 12,302; 22,487; Od.
2, 246; 18,318.
14) 'Setting the scene' may be seen as the general discourse-function of the sentence-ini
tial conditional. The first discussion in which conditionals are described as the 'topics' of their
main clause is Haiman (1978). Further testing and elaboration in Ford & Thompson (1986).
Some application to Dutch and Greek in Rijksbaron (1986). Notice that what in Haiman (1978)
is called the 'topic status' of conditionals (cf. Ickler (1981), where conditionals are discussed
under 'situational topics') has in Functional Grammar to be discussed in terms of 'Theme' (see
the definition of 'Theme' in Dik 1978:19: "the Theme specifies the universe of discourse with
respect to which the subsequent predication is presented as relevant"). In spite of the confident
tone of the proposals just mentioned, the discourse function of conditionals still deserves much
investigation. Especially the order of the subclause and the main clause and its impact on the dis
course function of the subclause - clause-final conditional cannot be 'topics' - needs much fur
ther research. This does not only apply to conditionals, but to other subclauses and peripheral
elements too. For an interesting discourse-oriented discussion of the linear ordering of sub
clauses and main clauses see Ramsey (1987).
15) Ex. (7) may be read generic, too, in spite of the fact that it applies to a particular situa
tion. See Ruijgh (1971: 726).
16) Extensive discussion in Ruijgh (1971: 719).
17) Ruijgh (1971:687,720) points out that te in digressive independent clauses {gar te, de
te, alla te) is, within a diachronic perspective, more recent than te in non-restrictive relative
clauses (hôs te, hos té). This observation is important in connection with assessment of the
diachronic status of ei per. See below, 6.3.3.18) To my mind, a more fundamental value of de (especially in diachronic space) is that
of topic marking. The opposition between the continuative and the adversative use of the par
ticle may then be accounted for, in a unifying way, in discourse-pragmatic terms: 'topic
continuity' (see Givón, ed. 1983) vs. 'topic shift'. From Ickler (1981), it appears to me that there
are striking similarities between de, especially as it is used in older Greek (Homer) and the Old-
Russian particle ze. The latter is described by Ickler (1977,1981) in terms of topic marking.
19) Unless we take and in the sense of 'and moreover'.
20) Actually, the extraction (left-dislocation) of a constituent which is marked by de is not
confined to concessive conditionals; it is a common phenomenon in Homer. Compare the fre
quent (formular) expression ton d'hôs oun enoêse ('And him(,) when he saw him') as well as
cases like Tudeidên d'ouk an gnoiês poteroisi meteiê ('And Tudeides(,) you could not discernamong whom he was', Il. 5,85). See also ex. (37) in 6.4.3 below. This kind of topicalization, which
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 245/318
NOTES TO CHAPTER 6235
in other terminology may be termed 'thematization' (see Dik (1978: 132 ff.), is highly charac
teristic of the Greek epic language, and less so for later Greek. It deserves of an investigation in
its own right, preferably within the perspective of orality and the influence of writing and literacy
on language use.
21) Cf. Thompson & Longacre (1985: 197) and König (1986: 233). See also ex. (23) in
2.3.1.6 above.
22) See also Ruijgh (1971:829). As regards 'rhetoric' Od. 20,49 is comparable to this ex
ample.
23) Notice that sometimes factuality (viz. specific reference to a particular, existing state
of affairs) is absent anyway, because the proposition functioning as conditional is generic, i.edevoid of any specific spatial or temporal reference (see 2.1.4 above). This is the case in (8)
(simile), (9)), (10), (11) (simile) and (12) (simile).
24) Rieu translates "if you care about your family at all", but the negative polarity item at
all does not capture the scalar properties of the expression. In later Greek, there are many in
stances of ei per which involve the semantics of at all. In the present study, however, we cannot
discuss these (see however ex. (37) in 6.4.3 below with note 53). See Bakker (in prep.).
25) Notice that in Dutch self ('zelf') and even ('zelfs') are etymologically cognate. Further
more, the relevant expressions in Romance (Fr. même, It. medesimo, Sp. mismo: 'even') are
derived from Lat. memet/temet/semet ipsimum ('me/you/him self ). I owe the last observation to
prof. C.J. Ruijgh.
26) After Homer it is even obligatory, yielding a separate conditional subordinator, ean( = ei + an) beside ei.
27) According to Chantraine (1953:278). Notice, however, that ke(n), the Aeolic counter
part of an, is by no means rare in optative conditionals in Homer, see Monro ( 1891:285-86). In
these cases, ke seems to have its original function as a demonstrative: 'then', 'in that case'. For
an example, see ex. (38) in 6.4.3 below.
28) See Kühner-Gerth (1904: 482), Goodwin (1889:147,192). 'Optative + an' ina con
ditional is an example of a typical main clause phenomenon ('potential optative') occurring in a
subordinate clause. The conditional containing 'optative + an' is, paradoxically, an indicative
conditional ('if it is true that..').
29) Compare . 5,232, some lines later, where eiper an aute is localized in 5.5, before thetrochaic caesura. This is of course in line with the localization of the - ⋃⋃ - ⋃ word-type (cp.
the localization of akhnumenosper in 5.2.2 sub I.A1) above).
30) In the terminology of Nagler (1967: 281), eiper an autos /tai and ei per an aute could
be characterized as two allomorphs of one and the same preverbal Gestalt in the poet's mind.
Another instance is the phrase hêi per an houtos ( Il . 7,286).
31) It is interesting to note that the ei per an autos/aute formula exclusively occurs in the
first five books of the Iliad (both instances of ei per an aute even occur in close proximity: Il. 5,
224; 232). This is an example of what Ruijgh (1957:20-1) calls "agglomération". The agglomera
tion of expressions which are similar in sound and/or meaning is a phenomenon the frequency
of which is characteristic of oral poetry.
32) For the order of conditionals and their main clauses, see Ramsey (1987). Ramsey argues (and computes) that preposed conditionals (as well as temporals) have a much larger dis-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 246/318
236 LINGUISTICS AND FORMULAS IN HOMER
course-scope than postposed ones. The latter tend to refer back solely to their main clauses,
whereas the former tend to have more and wider connections in the discourse.
33) In the following subsections only examples of kai ei (ei kai) are given; it goes without
saying that what is said about kai ei equally applies to oud' ei, because oud' ei is simply a conces
sive conditional whose main clause is negative (see on scalarity/concession and negation 2.4.1,
3.2.2 (note 17) and 4.3 (note 40). For the instances of oud' ei in Homer (as well as elsewhere)
see Oguse (1968:267-271).
34) The instances of oud' ei are not included in this figure.
35) The other two factual instances: Il 3,215; 15,51.
36) For this particle in connection with the concept of 'redundant semantics' see 5.2.2 subA3) and B1) above.
37) The same T2-formula in Od. 8,139.
38) The same P2-formula, with a minor difference due to inflection in Od. 7,194.
39) But for (30) such an expression is hard to find. In Il. 15, 30 we have skai polla per
athlêsanta, but, firstly, this expression is not a regular formula; it belongs to the isolated and uni
que (from the point of view of metrics and versification) cases of 'participle + per', and second
ly (more importantly), the phrase has no scalar-superlative meaning, while in the context of (30)
a concessive expression which involves superlativeness is more suited than an expression which
does not.
40) In Rijksbaron (1984:76), on the other hand, it is claimed that in Classical Greek there
is no marked difference between kai ei and ei kai.
41) One of them is ex. (25); the other is. Il. 17,421.
42) Notice that oud' ei has the same metrical form as ei kai] accordingly, as a formula it
behaves in like fashion, as a P2-formula. Because ei and oud'(e) cannot be reversed, like ei and
kai, we do not find T2-formulas. But oud' ei can perfectly be localized in other places in the verse,
for instance in 1 (beginning of the verse).
43) See Shipp ( 1972). For a detailed account of the linguistic structure of the Homeric
simile see Ruijgh (1971: 846 ff.).
44) Most of the instances of adverbial te are to be found in similes, see Ruijgh (1971: 2,
25,846).
45) See exx. (11), (12) and (19), as well as Il 7.11,116; 12,302; 16,263.
46) There is one instance of ei per (eanper) which is, according to Kühner-Gerth (1904:
170) to be translated as 'selbst wenn': Plat. Euthyphr. 4B 10. But Denniston (21954: 488) rightly
states that this example has to be interpreted otherwise. See Bakker (in prep.).
47) Notice that -pote is semantically cognate to English -ever. Both may also be used in
negative polar contexts (ever, (pô)pote).
48) Another expression-type in Greek which may be used to express indeterminacy is the
subjunctive mood in a relative clause (preferably introduced by hosos). As an early example of
this usage Il. 3,66 may be adduced.
49) The rationale behind the neutralization is self-evident in the case of (34): when a scalar
term occurs in the conditional, the concessive marking of the conditional is optional; it just mayor may not occur. In the case of (35) the neutralization, viz. the interpretation of if as even if, is
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 247/318
NOTES TO CHAPTER 6 237
due to Gricean principles of cooperative conversation, as König (1986: 238) plausibly states: aquestion in the format of q if ? makes sense only if both speaker and addressee assume that ρis an unfavorable condition for q (hence the concessive interpretation), for if ρ would befavorable or even neutral, it would be useless to ask the question. As for (36), here the concessive interpretation seems to be triggered by the non-obtainance of the state of affairs referredto in the main clause and the factual nature of the conditional.
50) See Denniston (21954: 487-8), Kühner-Gerth (1904:112), Iiddell & Scott, s.v.
51) This is what Haiman (1983) calls the 'iconichy of language, the principle of 'one form- one meaning'.
52) Notice that between two senses there is no formal difference: they belong to one andthe same lexeme. Accordingly, we need not speak here in terms of the neutralization of the (formal) differences between two items (categories), but in terms of overlap: the set of instances ofa given lexeme to which sense 'A' applies overlaps with (has an intersection with) the set of instances (of the same lexeme) to which sense B' applies.
53) It might be objected here that the speaker is a seer, who in that quality already knowswhat will happen, whereas scepsis is a matter of uncertainty. But this objection does no justiceto Teiresias' communicative intentions: his point is to stress the terrible things that will happenwhen Odysseus and his men slaughter the cattle of Helios.
54) We might say that the conditional in (37) contains the semantic presupposition of themain clause in a hypothesized form. lSemantic presupposition' we call what a sentence implies
irrespectively of its being 'true' or 'false' (see Levinson 1983:174-75) and irrespectively of its discourse function. Thus the presupposition of He regretted that he had murdered Jones is He mur dered Jones. Likewise, whether or not Odysseus comes home in evil plight, he has escaped in anycase. Now semantic presuppositions may be 'suspended', when they enter in a conditional clause.The speaker then wants to explicitly state that the truth of the presupposition is not a matter ofcourse, as it normally is. This happens in (37). Presupposition suspension is a more usual function of ei per in later, post-Homeric times. Ei per then invariably is placed after its main clause,like most presupposition-suspending conditionals. Notice that in English, presupposition-suspending conditionals tend to contain Negative Polarity Items (, at all, compare überhauptin German). See in detail Bakker (in prep: ch. 7). See also Horn (1972: ch. 1).
55) Notice that this example constitutes a case of anaphoric ke in an optative conditional
(see on ex. (18) in 6.3.1.5 above).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 248/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 249/318
7 PERIPHERAL INSTANCES
7.0 IntroductionWe have now discussed the instances of per that belong to the recognizable
expression-types. In the present chapter we go on to discuss the remainingmaterial in the data-base. These instances form a heterogeneous group, butmost of them have in common that the presence oiper cannot be straightforwardly accounted for within the scalar-concessive framework. They are un-classifiable and/or divergent by their form or by their meaning. A number of
alternative ways are open to deal with these instances. It is with these alternative approaches that we are concerned in the present chapter.
It is important to notice that the distribution of the examples over thevarious sections and subsections of this chapter does not reflect any clear-cutdivision in the material: some instances can be discussed under more than oneheading. Rather, the division of the present chapter serves expository purposes;it is meant to provide the various points of view from which the 'residue' whichis left in the data-base can be considered.
7.1 Linguistic irregularity in Homer7.1.1 Unclassifiable vs. divergent instances
Any description of the actual use of a linguistic item has to allow for thefact that instances of the item which conform without problems to the definition exist side by side with cases to which the definition applies to a lesser degree. To acknowledge this fact is essential for any sane descriptive linguistics.Linguistic phenomena often do not allow of exact classification and seem tothwart attempts to arrive at a typology which is based on discrete kinds. In otherwords, an adequate description has to make a distinction between prototypical
and peripheral instances of the item described. This distinction is the core of
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 250/318
240 LINGUISTICS AND FORMULAS IN HOMER
Prototype Theory, which was introduced in 1.4 above. A second element of thistheory is the notion of continuous space between two extremes. This notion has
played an important part, too, at various points of the argument of the presentstudy.
Peripherality has a number of aspects, which have to be kept distinct. Agiven instance of per may be called peripheral either when it is unclassifiableor when it is divergent from a norm. Unclassifiable instances of per are the onesthat do not belong to one of the recognizable expression-types discussed in
chapters 3-6. However, this isolated status need not preclude them from beingfull-fledged scalar particles. In 3.6 above I have presented some unclassifiableyet scalar instances. The willingness to call them 'peripheral' depends on thevalue one attaches to the classification into various expression-types: if one is prepared to claim that the classification reflects a real difference for the contemporary language user, the instances may easily acquire the status of exceptions. On the other hand, if the classification into expression-types is merelyused for the sake of presentation, the unclassifiable instances may not be'peripheral' at all. However, there are also unclassifiable instances which cannot possibly be called scalar; to these the predicate 'peripheral' does apply.
They will be discussed in 7.5 below.Peripherality may also involve divergence. The norm with respect to which
an instance of per may be divergent is one of the expression-types discussed inchapters 3,4 and 6. Instances of per may be divergent either as to their meaning or as to their form. In the first case, the (scalar) meaning undergoes a change(viz. disappears), while in the second it is the specific form of the expression-type that changes. 'Divergence' will be discussed in 7.3. In 7.3.3 I shall arguethat a change in form is always concomitant with a decrease in (semantic)
prototypicality.
7.1.2 The explanation of peripheral instances in Homer
The existence of peripheral instances which form the residue of the data base is characteristic of the description of natural language facts in general, butit applies a fortiori to 'Homeric' linguistics. In Homer, the peripheral instancesof a given item are more numerous than in ordinary language. Not all of thoseinstances are peripheral without more ado; a number of them can be explainedon the basis of some explanatory parameters that are specifically 'Homeric'.The nature of the Homeric poems, as well as the way they are composed, is
such that pure linguistics cannot, and need not, cover all the instances of a givenitem. The residue left does not only comprise those instances that are diver-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 251/318
PERIPHERAL INSTANCES 241
gent and/or unclassifiable in linguistic terms. Homeric linguistics has to allow,in addition, for the existence of peripheral instances the analysis of which transgresses the boundaries of linguistics proper.
The residue of a given item in Homer is a heterogeneous affair. In the peripheral status of a given instance, either of the following three factors may be involved.
1) Redundant semantics. A given linguistic item may be used to changethe metrical form of a given phrase or expression, so as to adapt it to the exigencies of the verse. This practice yields instances of the item which have aredundant and hence peripheral status. A necessary condition for this usage inthe case of particles (e.g. kai in participial phrases, see 5.2.2) is that the'redundant' element is possible anyway in the language, as a 'strengthener' (see4.4.2 above).
2) Semantic integration of formulas. Peripherality may be due to the waythe formula in which an item occurs is actually used in the epic diction. A lessthan perfect semantic integration of the formula may yield a less than prototypical or even peripheral status of the linguistic item(s) of which the formula con
sists (see also 5.3.2).3) Solution of ad hoc metrical problems. This factor differs from the other
two in that no specific formular connections are involved. Small words with ahigh frequency in the language are apt to be used to prevent hiatus or create along syllable which is metrically necessary. The resulting redundance is highlycharacteristic of Homeric poetry. After all, the metrical space of the hexameterhas to be filled, and rapidly filled, also when there are no formulas or formular
phrases at hand.
It is important to realize that these three factors do not create any clear-
cut tripartition in the material (the peripheral instances of a given item). Veryoften it is impossible to assign a given peripheral instance exclusively to one ofthe three. Rather, the three factors together constitute a general frameworkwithin which the peripheral instances of a given item can be assessed.
Beside the internal constituency of the framework formed by the threefactors, there is the question how far the framework extends. In other words,when is a given peripheral instance to be accounted for within the frameworkof epic diction and when is it not? This is a difficult question and one whichdoes not allow of a clear answer, as we do not possess any contemporaneous
linguistic material in the form of (non-formulaic) prose texts. It may seem attractive to treat a problematical instance of a given item as a stop-gap and to
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 252/318
242 LINGUISTICS AND FORMULAS IN HOMERadduce the third factor for its explanation. But it is very hard to determinewhether it is not a normal peripheral case which would still be present if metrical pressure were removed. And to apply the concept of semantic integrationof formulas may seem the appropriate solution to some problematical instanceof a given recurrent phrase, but, again, it is impossible to determine whetherthe boundaries of ordinary language have been transgressed. As for the firstgroup of the Homeric residue, the redundancy here amounts to a mere sys-tematization of what is present anyway in ordinary language.
Generally, we may say that in Homer, in spite of the constraints and exigencies of the oral-formulaic 'medium', the occurrence of expressions that arein overt and provable conflict with the Greek language as we know it, or thatwould not have occurred outside the epic diction, is very rare. This greatly adds,of course, to the difficulty to isolate the specifically Homeric part of the residue.
Sometimes there is yet another factor at work which transgresses the boundaries of linguistics proper. A given instance of per may have the normalscalar and/or concessive meaning of per but this does not appear from the ex
pression itself: the instance has properties that preclude it from being treatedas a prototypical instance. To account for the scalarity of such instances, wehave to appeal to what the poet might have meant in using the phrase. Such anattempt to justify the presence of a given instance of per means leaving therealm of linguistics proper and enter the domain of literary interpretation. Torecognize this possibility may seem strange in a linguistic study, but we have toallow for the possibility that a poet (or any other language user) may mould histhought in such an unusual form that the expression falls outside the scope ofthe description of the expression-type in question.
The examples in the present chapter will be presented according to thefollowing ordering principle. First (7.2), I will discuss the instances that recur
elsewhere in Homer. After these verbatim repeated phrases, I will discuss instances whose expression-type recurs elsewhere (7.3). These are the instancesthat are divergent from an expression-type either by their form or by theirmeaning. In the remaining sections, I will discuss unclassifiable instances bydegree of increasing peripherality. The examples in 7.4 are not scalar anymore.Accordingly, they lack the major prototypical property of per. However, theystill have a strong connection with inclusive focus particles, i.e. with the widersemantic framework within which scalar particles have to be discussed (see.2.3).In 7.5, finally, there is no question of there being any alternatives for the 'focusconstituent' of per. Here/per is most peripheral with respect to the prototypical scalar instances. In 7.61 will argue that there is a connection between a num-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 253/318
PERIPHERAL INSTANCES 243
ber of peripheral cases and the use of per in non-scalar and non-concessive subclauses which is the standard in later, post-Homeric Greek. This transitionmarks the boundary of the semantic field of 'Homeric' per as well as the end ofthe present investigation.
7.2 Semantic integration of formulas again
Some of the peripheral instances to be discussed in the present chapterdiffer from the others in that the phrase in question recurs elsewhere, in a
semantic environment in which no problem arises as to the function and meaning of per. In these cases it is possible to opt for the approach in terms of semantic integration of formulas.
Consider the following example:
(1) (Thersites, the ugliest of the Greeks before Troy, and a notoriousquibbler, is making profit of a general disorder in the Greek army.He abuses Agamemnon and tries to persuade the Greeks to sailaway, instead of enduring hardships before Troy for the benefit ofthis man:)ô pepones, 'elengkhe', Akhaïdes, ouket'Akhaioi,oikade per sun nêusi neômetha, tonde d'eômenautou eni Troiêi gerapessemen."As for you, my friends, poor specimens that you are, Achaeanwomen -I cannot call you men - let us sail home by all means andleave this fellow here to batten on his spoils." ( Il.2,236).
The occurrence of per here, which takes as its focus constituent oikade ('home') poses a problem. Denniston (21954: 482) Hsts the example under the heading
'determinative' (use), citing Monro's (
2
1891: 320) rendering: "Let us havenothing short of return home." Evidently, we are supposed to conceive of peras lending emphasis to oikade. But mere 'emphasis' is unsatisfactory as an account of per (or any other (Greek) particle, for that matter), especially when itis meant to apply to the prototypical uses of the particle. When dealing withthe prototypical uses, we can and should be much more specific.
If we are to speak about a given meaningful instance of per in terms of'emphasis', this must be the emphasis of the focus constituent of a scalar focus
particle. But in the present case there is nothing scalar: we have here a simpleadhortation ('Let's go home'), in which the focus constituent 'home' does not
include alternative values. At best we may say that oikade ('home') is contrasted
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 254/318
244 LINGUISTICS AND FORMULAS IN HOMER
with autou eni Troiêi ('here in the land of Troy'). But for the linguistic realization of such a contrast, per is strange.
Thus we have here what seems a definitely non-scalar instance of per;more will follow in the present chapter. As part of the solution to the problem
posed by per in (1), we may adduce the fact that the phrase and adhortation inquestion recurs elsewhere in the Iliad. Consider:
(2) (When Patroclus is ready to join battle, dressed in Achilles' armour,Achilles addresses his men, the Myrmidons as follows: "Often youused to say to me: 'You are a brute, Achilles, for keeping us out of
battle against our will'":)oikade per sun nêusi neômetha pontoporoisiautis, epei rha toi hôde kakos kholos empese thumôi"Let us then at least return home with our seacleaving ships. For youare in a too evil mood." ( Il. 16,205).
As to the form we have here the same adhortation as in (1), but here it is scalar.The use of per in (2) belongs to the expression-type described in 3.5 above, theuse oiper in wishes and commands that is associated with desirability scales. Fora Myrmidon warrior, the best thing you can wish for is fighting. But fighting isimpossible under the given circumstances, because of the wrath of Achilles andhis refusal to join battle with his contingent. So the Myrmidons have to resortto a compromise, viz. the highest item but one on their desirability scale: sailing home (lying in idleness before Troy being the worst thing imaginable). Theconstituent 'home' {oikade) is the highest item on the satisfactoriness scale (see3.5.3) of the Myrmidons: the least thing desirable that still satisfies their wish.
Thus the occurrence of per in (2) can be accounted for in scalar, that is, inlinguistic, terms. But then (1) is accounted for, too, so it seems, by way of the
concept of semantic integration of formulas. We may reason that for formularreasons the poet has used a scalar adhortation, which is, unfortunately, seman-tically out of place. The context of (1) has nothing to do with desirability scalesand second best possibilities. But that does not matter. Formular pressuresimply is greater than the desire to achieve semantic well-formedness.
The approach in terms of formular integration is useful in the case ofoikade per in (1), but it cannot go all the way, I think, to a complete solution:the semantic distance between the normal and the badly integrated case simply is too great. In other cases of less than perfect semantic integration (for example agathos per eon in ex. (65) in 5.3.2), per is less than prototypical (see4.3.4), but the phrase does remain an instance of concessive 'participle + per'.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 255/318
PERIPHERAL INSTANCES245
In the case of oikade per in (1), on the other hand, all that remains of the scalarityof the proper use in (2) is some vague sense of emphasis. In other words, itmight be asked whether the 'catachrestic' use of oikade per would have been
possible, even in the formulaic diction, if there had not been a possibility to use per anyway as some vague emphasizing particle, irrespectively of its occurrencein the formula oikade per. In view of the examples to be presented further inthis chapter (especially those in 7.5), I am inclined to answer this answer with'no'.
The following example discussed here brings us to the last part of the Odyssey:
(3) (Penelope is telling Odysseus (who is in disguise) a dream she hashad: she has twenty geese which she delights in watching. Suddenlyan eagle appears which kills them all:)autor egô klaion kai ekôkuon en per oneirôi"I wept and cried aloud in my dream." (Od. 19,541).
The prepositional phrase in this example, which functions as a bucoUc clausula,
raises problems. A possible interpretation of it, which does at least justice tothe presence of per, is concessive. However, when we interpret (3) as 'I weptand cried aloud though it was only a dream,3 who is weeping? The Penelopewho is dreaming, or the Penelope in the dream? The former possibility is ruledout, for the dream goes on. But the second possibility makes a concessive interpretation of en per oneirôi strained: a person who is in a dream knows she isin a dream and conceives of her crying as something she is doing in spite of her
being in a dream.A better way of dealing with en per oneirôi in (3) is leaving interpretation,
whether or not subtle, aside for a moment and look at the recurrences of the phrase in question:
(4) (Somewhat later in book 19, Penelope tells the beggar the strategyshe has chosen: she will hold a bow contest. "Whichever of thesuitors proves the handiest at stringing the bow and shoots an arrowthrough each of the twelve axes, with that man I will go:")
nosphissamene tode domakouridion, mala kalon, enipleion biotoio,tou pote memnêsesthai oïomai en per oneirôi
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 256/318
246 LINGUISTICS AND FORMULAS IN HOMER"Bidding goodbye to this house that welcomed me as a bride, thislovely house so full of good things, this home which even in mydreams I never shall forget." Od. 19,581).4
Here en per oneirôi has a clear and unambiguous meaning which precludes anyuncertainty about its interpretation. The phrase functions as a pragmatic su perlative in a non-factual (futural) environment (see 2.1.3 and 2.1.4 above): 'tillin my dreams'. The phrase evidently represents the high point of a scale: someone who remembers something in his/her dreams will a fortiori do so whileawake. Like all other statements containing even or a scalar superlative, therelative clause in (4) has a well-defined illocution too: the assertion of a remark-able and superlative fact. In (4) this is of course Penelope's enormous attachment to Odysseus' house.
It seems to me that the linguistically unproblematic occurrence of en peroneirôi in (4) is a factor to be reckoned with in the interpretation of (3). Bothinstances of the phrase not only occur in close proximity of one another; theyalso belong to one and the same episode: both are part of Penelope's wordsconcerned with her point of view with regard to the Suitors. En per oneirôi is a
formula with a limited range of occurrence: as far as we can see on the basisof extant Greek epic, it is confined to one particular theme which occupies awell-defined place and role in the narrative. Consequently, the imperfect integration of the formula occurs near to the perfect one.
However, just as in the case of kade per in (1), I do not believe that en per oneirôi in (3) should be accounted for in purely formulaic terms. The concept of semantic integration is just a factor not to be neglected. But the non-scalar use of per in (3) may well belong to the peripheral instances of per, inwhich scalarity as the major prototypical feature has disappeared.
7.2.1 Formular or linguistic irregularity?
Beside (1) and (3), there are more phrases featuring per that are non-scalar and that recur elsewhere in a scalar form. For these examples, however,the approach in terms of formular integration, though it is still possible, is lesssuitable: in one case (ex. (5) below) the alleged badly integrated formula isrepeated in its own right, and in another case (ex. (8) below) it can hardly becalled a formula at all. These two examples may serve, in the present exposition, as transitional cases between exx. (1) and (3) above and the examples to
be presented below (7.5).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 257/318
PERIPHERAL INSTANCES 247
Consider:
(5) In book 22 of the Odyssey it is told how, after the murder of theSuitors, the unfaithful maidservants are collectively hanged:êspairon de podessi minuntha per, ou ti mala dên."They kicked with their feet for a little while, not very long." (Od.22,473).
Again, the occurrence of per, which here modifies a temporal adjunct in a fac
tual context, is problematic. Not only is per non-scalar; it is not inclusive either,as is shown by the bucolic clausula which follows (ou ti mada den, 'not verylong'). The phrase minuntha per, ou ti mala den in a factual environment recursa couple of times ( Il. 1,416; 13,573).
The three examples belong to the cases where an interpretation in termsof the supposed etymology of per (see 1.3.4) has been proposed ('a very shorttime'). Accordingly, the use of per exemplified by (5) is considered to be veryold in this approach. But to suppose that (5) exemplifies an old relict, whichhas otherwise disappeared from the language, is entirely gratuite. To my mind
there is no reason whatever to assume that the use of per in (5) is particularly'old' . Just as all the other alleged instances of the supposed 'origina'('intensive') meaning of per 8 , it can be accounted for otherwise.
The phrase minuntha per as it occurs in (5) occupies a central place withinthe present chapter; it can in principle be discussed both in connection with thesemantic integration of formulas and in connection with peripherally (see 7.5
below), thereby indicating the overlap between these concepts.The use of per in (5), where it fills, together with minuntha, the slot be
tween the T-caesura and the bucolic diaeresis, can in principle be analyzed asthe imperfect semantic integration oiminunthaper as it occurs in adhortations
and commands. Minunthaper in these contexts entirely conforms to the analysisin terms of desirability and satisfactoriness scales of section 3.5.3, just as oikade
per in (2). There are three instances of minuntha per in adhortations ('if onlyfor a little while'). One of them has already been presented as ex. (37) of ch. 3;the other two are presented here:
(6) (Menestheus sends a herald to both 'Aiantes':) Aiant'y Argeiôn hêgêtore khalkokhitônôn,ênôgei Peteôo diotrepheos philos huioskeis'imen, ophra ponoio minuntha per antiasêton,
amphoterô men mallon.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 258/318
248 LINGUISTICS AND FORMULAS IN HOMER
"My lords and commanders, (he said), Menestheus my noble master begs you to come over, if only for a little while, and lend him a handin his difficulties, preferably both of you." (77.12,356).
(7) (In his dream Achilles addresses the ghost of Patroclus:)alla moi asson stêthi, minimtha per, amphibalonteallêlous, olooio tetarpomestha gooio."But now come nearer to me, so that we may hold each other in ourarms, if only for a moment, and draw comfort from our tears." (/7.23, 97).
In these examples, we see minuntha per in a semantic environment in which per can be analyzed as a full-fledged scalar particle. The realis environment in(5), on the other hand, robs per of its proper scalar meaning. However, thereis reason to refrain from an approach to minunthaper in (5) in terms of semantic formular integration. Minunthaper as it is used in (5), followed by ou ti maladên, is apparently a formulaic expression in its own right: it occurs three times.This makes the expression less suitable for an analysis in terms of semantic integration: for the very point of this approach is that the badly integrated instance is not a formula in its own right, but a 'catachresis' of it.
We have to allow for the possibility, then, that minunthaper in (5) is simply a highly peripheral and atypical instance of per which is not connected, by
being a formula, with the scalar per in (6) and (7). Indeed, per in (5) is so atypical and unlike any of the other instances of the particle (except the ones
presented in 7.5 below) that we have to admit that the difference with the particle ge is minimal. We might even consider the possibility that ge was not usedin (5) solely because it is prosodically infelicitous (a consonant is needed beforeou ti mala dên).11
The following two instances of the combimtion gunaikiper (Vornan', dat.)are in more than one way comparable to the instances of minunthaper above:
(8) (Thetis to her son Achilles, who is still mourning for Patroclos:)teknon emon, teo mekhris oduromenos kai akheuônsên edeai kradiên, memnêmenos oute ti sitououf eunês? agathon de gunaiki per enphïlotêtimisgesth'ai)."My child, how much longer are you going to eat your heart out in
lamentation and misery, forgetful even of your food and bed? It isgood for you to make love to a woman." (i7.24,130).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 259/318
PERIPHERAL INSTANCES 249
(9) (The soul of Agamemnon to Odysseus, after he has told how he wasmurdered by his wife Clytaemnestra:)toi nun mê pote kai su gunaiki per êpios einai."Let this be a lesson to you also. Never be too gentle even with yourwife." {Od. 11,441).
Although per in (9) certainly does not belong to the most characteristic instances of the particle, it allows at least of a scalar reading. A scalar or even an inclusive reading of per in (8), on the other hand, has curious consequences: 'Itis good for you to make love even/also to a woman'. But homosexuality is soalien to Achilles and the Iliad as a whole that the intended meaning of gunaiki
per in (8) can hardly be scalar. Per in (8) is just as peripheral and atypical asin (5), and just as in the case of minuntha per, it is false to speak oîgunaikiperin (8) and (9) as the imperfectly resp. the perfectly integrated case of one andthe same formula.
73 Divergence: changes in form and meaning
In the present section, I discuss instances that are either by their form or
by their meaning reminiscent of scalar expressions discussed in chapter 3 and4. In subsection 7.3.1,I present some examples which have retained the originalform of the expression-type but lost the scalar meaning, and in 7.3.21 will discuss examples which have retained their meaning but have changed their form.Strictly speaking, the non-scalar instances that have retained the form of theoriginal expression-type are more peripheral from a semantic point of viewthan those which still have a scalar meaning. But for the sake of parallelism
both groups are discussed as two subsections of one and the same section.
7.3.1 Changes in meaningSometimes we find instances of per which belong as to their form to one
of the expression-types discussed earlier. However, they are not scalar. Consider:
(10) (Phoenix tries to persuade Achilles:)all'Akhilleu pore kai su Dios kourêisin hepesthaitimen, hê t'al lôn per epignamptei noon esthlôn."Achilles, you must give honour to the Daughters of Zeus ( = the'Prayers'), (honour), which bends the mind of other sensible
people." (77. 9, 514).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 260/318
250 LINGUISTICS AND FORMULAS IN HOMER
The original expression-type here is 'adjective + per' functioning as a scalarsuperlative in a generic hos te clause (see 3.1.1). In (10) the characteristic hoste-clause contains the combination allôn per. On account of the properties ofalios ('other'), this combination cannot function as a scalar expression: other(than) is exclusive. Thus the overall structure of the original expressionremains, while the meaning has changed.
Notice that the discourse-function of the expression has changed, too. Theoriginal expression, where 'adjective + per' in a hos te-clause has the meaning
of a scalar superlative, functions in accordance with the illocutionary potentialof sentences containing a scalar expression (see 2.1.2 and 2.3.1.2), to state a su perlative fact. This statement is invariably made on the part of the narrator ina digressive, backgrounded statement (see 3.1.3 and 3.2.4). In (10), on the otherhand, we have a directive speech act (see 3.2.2) performed by one of the characters: Phoenix is trying to persuade Achilles to give up his anger. On account ofalios, however, his argument does not have the a fortiori character of the instances of kai discussed in 3.2.2.
Ex. (10) occurs, interestingly from the point of view of oral composition,in close proximity of the following example:
(11) α ll' hote dê Meleagron edu kholos, hos te kai allôn
oidanei en stêthessi noon puka per phroneontôn.
"But then Meleager was overmastered by passion, and that is whathappens to other sensible men too." (Il. 9,554).
Here alios occurs also. But here it is the focus constituent of a non-scalar instance of kai, not of per. Per instead functions to modify an adjectival combination which recurs elsewhere (see also ex. (2) in 3.1.1):
(12) (The girdle of Aphrodite:)enth'eni men philotês, en d'himeros, en d'oaristus parphasis, hê t' eklepse noon puka per phroneontôn.
"It contains Love and Desire and the sweet bewitching words that
delude the mind of the cleverest people." ( . 14,217).
This example is scalar, noon.puka.per phroneontôn functioning as a full-fledgedscalar superlative. Accordingly, the use of the phrase in (11), where it is a kindof apposition to the non-scalar (as well as non-directive) phrase kai allôn, may
be treated as a case of imperfect semantic integration of the formula in ques
tion. There is recurrence both on the level of expression-type and on the levelof actual wording.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 261/318
PERIPHERAL INSTANCES 251The following two examples are instances of per in comparative expres
sions. The original scalar meaning of per here (for which see 3.3) has disap peared, and the particle seems to be merely used because it is an organic partof a rhythmically pleasing bucoUc clausula:
(13) all' aiei te Dios kreissôn noos ê per andrôn."But the thoughts of Zeus are always stronger than the thoughts ofmen." ( Il. 16, 688).
(14) (Diomedes to Odysseus:)êtoi ego meneo hai tlêsomai: alla minunthahêmeôn essetai hêdos, epei nephelêgereta ZeusTrôsin de boletai dounai kratos ê per hêmin"Indeed I will stand and take what comes. But our friends will not
benefit by that for long. Zeus the Cloud-bearer has decided he wouldrather see the Trojans win than us." ( Il. 11,319).
Any attempt to read a scalar meaning in these examples is bound to fail: 'thethoughts of Zeus are stronger even than those of men' is certainly not the in
tended meaning of (13) and, besides, a definitely un-Homeric way of thinking/saying. In Homer, the thoughts of humans simply are very weak and thoseof gods very strong. Likewise, in (14) there is a straightforward, non-scalar op
position: Zeus wants to see the Trojans win, and not the Greeks; there is noquestion of his wanting to see the ones win even more than the others.
Finally, I present an instance of ei per (gar) in which the proper meaningof that subordinator is absent:
(15) (Hecuba is imploring her son Hector to come inside the walls andnot to fight Achilles:)skhetlios; ei per gar se kataktanêi, ou s'et' egôgeklausomai en lekheessi."He is a savage; and you need not think that, if he kills you, I shalllay you on a bier and weep for you (for he will let the nimble dogseat you)." ( Il. 22, 86).
The conditional is not concessive; nor is there any trace of the characteristicsceptic modality which is the natural outcome of the use of even if in directivespeech acts (see 6.4 above).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 262/318
252 LINGUISTICS AND FORMULAS IN HOMER
7.3.2 Changes inform
The examples in the previous subsection have retained their form andchanged (lost) their original scalar meaning. In the examples to be presentedin the present subsection, it is the other way around: they have retained thescalar meaning but have changed their form. The change in form is due to amovement of per from its original locus. Consider:
(16) espelagos mega toion, hothen te per oud' oiônoi
autoetes oikhneusin, epei mega te deinon te."Into that wide expanse of sea, which is so vast and perilous that even
the birds cannot make their passage in the year." (Od. 3,321).
Like exx. (10)-(11) above, this example is based upon the expression-type'scalar superlative in non-restrictive relative clause modified by te' Per has
been removed from its original place, forming a collocation with te} Anotherexample:
(17) (Menestheus sends a herald to both Aiantes (see (6) above): 'Askthem to come over':)ei de sphin kai keithiponos kai neikos orôren
alia per oios itô Telamônios alkimos Aias."But if the two of them are as hard-pressed as ourselves, let us at
least have the brave Telamonian Aias." ( Il. 12,349).
Here the basic expression-type is the use of per in wishes and commands (see3.5). Again, per has moved away from its original locus: properly speaking, oios('alone') is its focus constituent. In (17), per modifies the whole command instead, but its meaning stays the same.18
In the rest of the examples to be discussed in the present section, the scalarterm oude is involved. The examples differ from (16)-(17) in that the formalchange is concomitant with a decrease in prototypicahty, whereas with per in(16)-(17) nothing is amiss semantically. Consider as a start:
(18) (Zeus wants to make love to Here. He assures her that they will not be seen:)- ⋃⋃ oud' an nôï diadrakoi Eelios per
hou te kai oksutaton peletai phaos eishoraasthai."Even the Sun, whose rays provide him with the keenest sight in all
the world, will not see us through the mist." ( Il. 14,344).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 263/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 264/318
254 LINGUISTICS AND FORMULAS IN HOMERIn the case of this example the basic expression-type is 'kai hôs' followed by'participle + per' (see 4.2.1). Again, per has moved from its natural position,this time the participle, to the focus constituent of the scalar particle in question. Note that ex. (19) not only differs from the original examples as to its form;semantically there is also a difference, in that the analysis proposed in 4.2.1 forkai/oud' hos does not apply here: hôs cannot be conceived of as an anaphoric pronoun referring backwards to the circumstances in spite of which the stateof affairs denoted by the main clause obtains. Instead, oud' hos has the idiomatic
sense which is normal in later times (see 4.2.2), 'nevertheless', 'all the same'.Thus, concomitant with the formal difference we have in (19) a semantic difference with the original expression-type. However, the difference does not involve the loss of scalarity, as in (10) and (11) above. Rather, the change involvesthe normal development of the meaning of per, in which scalarity is convertedinto concession.
Exx. (18) and (19) may be seen as transitional cases between the originalexpression-type discussed in 3.2.2 and 4.2.1 and the examples to be discussed
presently. Per has been removed from its proper seat, but that proper seat, anon-restrictive relative clause or an apposed participle, is still present. The addition of per to êelios in (18) may be due to difficulties in the versification, butthe following examples show that apparently no boundaries of what is linguistically possible have been transgressed: the direct addition of per to the focusconstituent of oude apparently has become a practice in its own right, and theapposed element, the most characteristic feature of the original expression-type, and the proper seat of per, is lost:
(20) (Achilles is dragging the dying Lycaon to the river:)oud' humin potamos per eurroos, argurodinêsarkesei, hôi dê dêtha poleas hiereuete taurous."Nothing shall save you, not even fair Scamander of the SilverEddies, to whom for years you have been sacrificing bulls." ( Il. 21,130).
Admittedly, there is a (non-restrictive) relative clause here, but its meaning isnot nearly as scalar (superlative) as that of (18) and the examples in 3.2.2. Itmay in principle be interpreted as carrying a concessive tone, but this is not avery useful thing to say, for concession is not inherent in the meaning of therelative clause. The relative clause is absent anyway in the folllowing examples:
(21) (Odysseus tells about the departure of him and his comrades fromthe house of Circe:)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 265/318
PERIPHERAL INSTANCES 255
oude men oud' enthen per apêmonas êgon hetairous."But not even now did I get them all off without a casualty." (Od. 10,551).
(22) (Athena to Odysseus, after the latter has just landed on Ithaca:)skhetlie poikilomêta, dolôn at', ouk ar' emettes,oud' en sêei per eôn gaiêi, lêksein apataônmuthôn te klopiôn, hoi toi pedothen phïloi eisin.
"And so my stubborn friend, Odysseus the arch-deceiver, with hiscraving for intrigue, does not propose even in his own country todrop his sharp practice and the lying tales that he drops from the
bottom of his heart." (Od. 13,294).
The following examples show that the expression-type oude X per may beused as a formula. On account of its convenient metrical form (bucolic clausula)the phrase oude nu soi per ('not even to you') is useful as a formula; it allowsfor flection, in that the personal pronoun soi (dat.) may be replaced by ademonstrative element:
(23) (Athena to her father Zeus: 'Why don't we rescue Odysseus'?:)oude nu soi per
entrepetai philon êtor, Olumpie?"Even your heart is unmoved." (Od. 1,59).
(24) (Hector urges Melanippus to defend the body of his cousin Dolops:)houtô de, Melanippe, methêsomen? oude nu soi per
entrepetai philon êtor anepsiou ktamenoio?"Melanippus, are we to take things down like this? Is it nothing even
to you that they have killed your cousin?" (R. 15,553).
(25) (Poseidon to Apollo, when they meet on the battle-field:)nêputV hos anoun kradiên ekhes: oude nu ton per
memneai, hosa dê pathomen kaka Ili amphi."But what a fool you are and what a short memory you have! Youseem even to have forgotten all the hardships you and I endured atIlium [you do not even remember those things..]." ( Il. 21,441).
The following example represents the end-point of the development. Here per is not attached to the focus constituent of oude, but directly to oudeitself, forming the collocation oude per.22 It occurs only once in Homer:
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 266/318
256 LINGUISTICS AND FORMULAS IN HOMER
(26) (The soul of Agamemnon to Odysseus: 'You will reach your homeand you will see your wife and your son':)hê d' erne oude per huios eniplêsthênai akoitisophthalmoisin ease; paros de me pephne kai auton."Whereas that wife of mine refused me even the satisfaction ofsetting eyes on my son. She could not wait so long before she killedhis father." (Od . 11,452).
7.3.3 From prototype to peripheryAs to their form, the examples of oude {..) per discussed in the previous
subsection may be ordered by degree of increasing divergence from the norm,which is posed by the original expression-type:
(27) oude X, hos per oude Xper, hos oude Xper oude per Χ
The leftmost item on this 'scale' represents the original expression-type discussed in 3.2.2 above; the second item represents exx. (18) and (19) above; thethird item represents exx. (20)-(25) and the rightmost item ex. (26). As a general
rule, we may say that the more rightward an item has to be localized, the moreit is divergent as to its form from the original expression-type.
But a change in function and meaning is involved as well: only at the left part of the scale has per its normal function, which has to be stated in terms ofsuperlativeness and intensionality. In the right part of the scale, per is attachedto extensional terms and the connection with intensionality has disappeared.But wherever a given instance of oude (..) per has to be localized on the scale,one thing stays the same: the instance always involves scalarity. This is why theexamples in question have been dealt with under the heading 'changes of form',rather than 'changes in meaning'.
Thus the examples discussed in the previous subsection may be seen asdisruptions, more or less drastic, of the original expression-type leftmost onscale (27). Yet I do not believe that any of them is artificial in the sense that itcould not have occurred in the ordinary language, outside the epic diction. Toclaim this would indeed be an untenable position, in view of the occurrence ofthe collocation oude per after Homer, in non-formular texts. Rather, wewould have to say that increasing formal divergence from the norm, as dis played by (27), reflects decreasing prototypicality. In other words, the morerightward a given instance of oude (..) per has to be located on scale (27), the
less specific is the contribution that per makes to the meaning of the scalar ex pression. At the left extreme of scale (27), per has its own, clearly recognizable
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 267/318
PERIPHERAL INSTANCES 257meaning and function and is as such maximally differentiated from oude. Onthe right extreme, on the other hand, it might be asked about a given instanceof per why it occurs at all, as it has a minimal, and minimally specified, function.
Closely connected with prototypicality is predictability. In the leftmost partof the scale {oude X, hos per, discussed in 3.2.2), per is maximally predictable;its occurrence can be predicted on the basis of the description offered in 3.2.2:whenever the relevant superlative intension of the focus constituent of kai or
oude is added in the form of a non-restrictive relative clause, we predict thatthis relative clause is modified by per (hos per). When per is used this way, it
belongs to the prototypical instances of the particle; it has a clearly specifiedfunction in an expression-type which is easily recognizable as such. Consequently, it is a welcome item in the linguistic description of per, as prototypesare the most rewarding items to describe and predict.
In the rightmost part of scale (27), on the other hand, the occurrence of per is minimally predictable in linguistic terms. Per as used in exx. (20)-(26)is a typical case of strengthening (see 4.4.2). The distribution of per at this ex
treme should be dealt with in terms of the residue which is inevitable and concomitant with any linguistic rule or definition. The instances here are peripheral, rather than prototypical, and hence often 'redundant', strictlyspeaking. In the typically Homeric context of the hexameter, metrics may well
be an important factor in this distribution, but peripheral cases of a given linguistic item occur in ordinary language as well.
7.4 Per and inclusive focus particles
In the previous subsection I discussed instances of oude (..) per in which
per is clearly less than prototypical. In the present subsection I discuss instances that are even more removed from the prototypical instances. The instancesof per discussed in the present section are placed after the focus constituent ofkai: kai X per. Oude in the previous subsection is invariably scalar; kai in the present one is not. In most cases it is a simple (non-scalar) focus particle (see2.3): too or also. Thus, in the present subsection we are one important stepremoved from the original meaning of per: scalarity has disappeared.
In 5.2.2 the collocation 'kai + participle + per' was discussed. In this ex- pression-type per is, of course, the central semantic element, while kai is additional and a case of 'redundant semantics'. In the ordinary language, the
function of kai in participial phrases would be that of a strengthener. In the examples to be presented below, on the other hand, it is the other way around.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 268/318
258 LINGUISTICS AND FORMULAS IN HOMER
Here kai is the central semantic element which cannot be missed, while per isoptional and 'redundant'. In the absence of a cogent set of metrical constraints,the distribution of per here cannot be explained or the occurrence of individualcases predicted iin the same way as withkai in the participial phrase. Consider:
(28) (The Greeks are praying to Zeus just before the duel of Aias andHector: 'Grant Aias the victory':)ei de kai Hektora per phileeis kai kêdeai autouisên amphoteroisi biên kai kudos opasson."But if you love Hector too and wish him well, let neither man be
beaten and the fight be drawn." ( Il. 7,204).
(29) (Thootes speaking to both Aiantes, urging them to come over tohelp:)ei de kai enthade per polemos kai neikos orôren,alla per oios itô Telamônios alkimos Aias."But if there is hard fighting here too, he hoped that you at least, mygallant lord, son of Telamon, would join him." (77.12,361).
These examples are similar as to their form (ei de kai per) and as to theredundant status of per. In both examples, per is redundant, being optionallyattached to an instance of kai (non-scalar) and its focus constituent.25 In thecase of (29), the optionality of per is clearly brought out by the fact that the linein which it occurs (and indeed the whole passage) is arepetition of a just preceding passage. In the earlier version of the Hne (see also (17) above),
(30) ei de sphi kai keithi ponos kai neikos orôren"But if there is hard fighting there too..." ( Il. 12,348).
there are differences which are due to spatial deixis: the difference is between'you here' and 'they over there'. The occurrence of per in (29), in a line whichcontains adaptations, and as a means to effect the adaptations, is telling. Wehave here cases where the particle has lost much of its linguistic salience; it justmay or may not occur. Consequently, the use of per in (28) and (29) is minimal-ly predictable in a linguistic framework.
Some more examples:
(31) mûthoi de kai êôthen per esontai
Têlemakhôi kai emoi diaeipemen allêloisi
"In the morning too there will be stories for Telemachus and me totell each other" (Od. 4,214).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 269/318
PERIPHERAL INSTANCES 259
(32) (Agamemnon requests Menelaus to wake the Greek leaders: 'Calleach men by his father's name and let them have their dignities:')mede megalizeo thumôialia kai autoi per poneômetha."And do not be proud. Let's work ourselves too." ( Il.10,70).
In one case, the combination kai X per is scalar. Accordingly, the examplein question is similar to the instances of oude X per discussed in the previoussection:
(33) (Menelaus deliberating in a difficult situation:)ei de pou Aiantos ge boên agathoio puthoimên,amphô ' aûtis iontes epimnêsaimetha kharmêskai pros daimona per, ei pôs erusaimetha nekronPêleïdêi Akhillêï."And if only I knew where the great Aias could be found, the pair ofus might go and make another stand, even with Heaven against us,
and try to save the body for my lord Achilles." ( Il .17,104).
Notice the unusual case of brevis in longo in 5 after per. The poet evidently haddifficulties in filling the Pi-part of the line.
Occasionally, we find instances of kai X per that belong to the early(Homeric) cases of an expression-type which is very frequent in later Greek,
per in 'the same ay-contexts' (see ex. (15)-(16) in 1.2 as well as 2.3.2.2). I giveone example:
(34) (Hector praying for the future of his son:) Zeu, alli te theoi, dote de kai tonde genesthai, paid' emon, hôs kai ego per, ariprepea Trôessin."Zeus and the other gods, grant that this boy of mine may be, like
me, pre-eminent in Troy, as strong and brave as I." (Il. 6,477).
The normal expression in later Greek is hôsper kai egô ('just like me'). In thisexpression-type, it is kai that may or may not occur. In the present section, onthe other hand, we are concerned with peripheral and additional instances of
per, not of kai.
7.5 At the outer periphery
In this section we make the last step which leads from prototypicality to peripherally. The instances of 'oude (...) per discussed in 7.3.2 and 7.3.3 were
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 270/318
260 LINGUISTICS AND FORMULAS IN HOMER
less than prototypical, but they were scalar at least; the instances of kai (...)perin 7.4 were still less prototypical, but they were at least inclusive as a focus particle. The instances to be presented in the present section lack all of these
properties. They are least specified and least distinguishable from the neigh boring particle ge. Consider:
(35) (Nestor is repeating what Menoetius had said to his son Patroclus:'Achilles is of nobler birth than you and he is stronger, but you areolder:')
all eu hoi phasthai pukinon epos êd'hupothesthaikai hoi semainein: ho depeisetai eis agathon per."It is for you to give him sound advice, to set him an example and totake the lead, which he will follow to his own advantage." ( Il. 11,789).
(36) (Patroclus to Achilles: 'It is a shame that you are sitting idly by youship while the Greeks are being defeated':)ti seu alios onêsetai opsigonos per,
ai ke mê Argeioisin aeikea loigon amunêis."What will future generations have to thank you for, if you will nothelp the Argives in their direst need?" (Il. 16, 31).
(37) (Aeneas: ' It 's no good fighting Achilles: there is always a god who ishelping him':)ei de theos per
ison teineien polemou telos, ou ke mala rheanikêsei, oud' ei pankhalkeos eukhetai einai."However, if the gods decide to see fair play between us, he will not
have an easy victory, even though he likes to think that he is madeofbronze."(77.20,100).26
(38) (Odysseus to Achilles in the Assembly:)kreissôn eis emethen kai pherteros ouk oligon per
engkhei: ego de ke seio noêmati ge probaloimên polion, epei proteros genomên kaipleiona oida."You are a stronger man than I and not a little better with the spear,
but in view of my greater age and experience I may well claim tohave more judgment than you." ( Il. 19,217).27
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 271/318
PERIPHERAL INSTANCES 261
(39) (Aias the son of Oeleus to the great Aias: 'The man we just met wasnot Calchas our seer but one of the gods':)ikhnia gar metopisthe podôn êde knêmaônrhei' egnôn apiontos: arignôtoi de theoi per."His heels and the backs of his knees as he left us were proof enoughfor me - it is not hard to recognize a god." (7Z. 13,72).
We have arrived here at the bottom of the residue. In these cases themeaning of per is so unspecific that nothing is left of it but some vague sense of'emphasis'. And the emphasis is definitely not the kind of emphasis connectedwith scalarity or even of simple inclusive focus particles or focus contexts (see2.3.2). The examples are at the outer periphery of the semantic field of per ; theyseem to have minimal linguistic significance. This appears from the fact thatthe difference with ge is minimal. The vague boundaries here with ge, whichcould be firmly drawn in the case of other, more prototypical cases (viz. the useof per in wishes and commands, see 3.5.4 above) may seem an unsatisfactorysituation in the Hght of the previous chapters, in which per could be successfully delimited, both with regard to ge and with regard to kai. But then we have to
realize, again, that in principle any description of the actual use of a given itemin language which does not allow for exceptions and border-line cases has veryselectively explored its data-base: it applies only to the prototypical cases, ignoring the peripheral ones.
However, notwithstanding the principle of prototypicality, the semanticdistance between (35)-(39) and the prototypical, scalar cases of per is very considerable. It might even be asked whether it is the normal distance between agiven prototypical instance and a given peripheral instance of one and the sameitem. In view of two of the typical complications in Homeric linguistics, epicdiction and the exigencies of the verse on the one hand, and diachrony on theother, this question is highly relevant in the present connection. As for epic diction and its vehicle, the hexameter, per in (35)-(39) must have some value ofits own, for the poet evidently was under no metrical constraint to use per in asituation where ge would have been more appropriate: because per is placed atthe end of the line, ge could perfectly have been used instead. On the otherhand, in 4 of the 5 cases, exx. (35)-(38), per comes at the end of a bucoUcclausula. Accordingly, we have to take into account the possibility that per in(35)-(38) is merely used to fill the remaining metrical space at the end of arhythmically significant colon (see also ex. (13)-(14) above).
As to diachrony, the question is whether the semantic distance between(35)-(39) and the more prototypical instances is conceivable in one and the
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 272/318
262 LINGUISTICS AND FORMULAS IN HOMERsame synchronic framework, or whether we have to introduce the dimension'time' to account for it. This is a difficult question, because per in (35)-(39) isso unspecified that it does not provide any clue as to its diachronic status and
position. It cannot easily be placed on a diachronic scale with items placed before it, as is the case with more prototypical cases. In other words, it is hardto determine whether (35)-(39) are the product of some genuine diachronicdevelopment. Anyway, if there was such a development, it must have stood outside the observable main stream, in which scalarity gradually is transformed
into concession or some other recognizable semantic feature; the use of per exemplified by (35)-(39) does not occur after Homer. Perhaps it is a contem porary 'colloquial' use, which was avoided in the language of epic poetry.
7.6 From Homeric to Attic per
In a diachronic approach 'pointing backward' is not the only relation possible. A given item may also point forwards in diachronic space. An analysis interms of pointing forward, viz. to later, post-Homeric uses of per, seems to bethe right solution to a number of instances of per which cannot, just like (35)-
(39), be connected with any of the prototypical, scalar uses. The instances inquestion occur in an independent clause; yet they point forward to a use of perwhich is confined to subclauses. They involve the use of alios ('other'). All instances are, perhaps significantly, from the Odyssey. Consider:
(40) (Zeus to Hermes: 'Go to Calypso and tell her that she has to releaseOdysseus':)
Hermeia, su gar aûte ta t' alla per aggelos essi:numphêi eüplokamoi eipein nêmertea boulên."Hermes, in your capacity as our Envoy [lit: you are in all other
respects our messenger], convey our final decision to that dainty Nymph." (Od. 5,29)30
(41) (Telemachus to Peiraeus:)Peiraie Klutidê, su de moi ta per alla malista
peithêi emôn hetarôn, hoi moi Puloneis ham'heponto:kai nun moi ton kseinon agôn en domasi soisi
endulceos phileein kai tiemen, eis ho ken elthô."Peiraeus son of Clytius, of all who joined me on this trip to Pylus Ihave always found you the most reliable [lit: you are in every other
respect..]. Will you oblige me [also] now by taking charge of this
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 273/318
PERIPHERAL INSTANCES 263
guest of ours and treating him with every kindness and attention inyour own house till I come back?" (Od. 15,540).31
Just as in (35)-(39), per is far from prototypical here from the point of view ofscalarity. Its function is to lend emphasis to alios in the same vague way as in(35)-(39). Yet this time the expression-type to which per belongs is characteristic and clearly recognizable. Ex. (40) and (41) seem to be early instancesof the expression-type ta te alla kai X ('the other things as well as X') which isfrequent in classical Greek prose. Its meaning is 'especially (above all) X',whereby ta te alla is virtually an idiomatic expression: its meaning cannot bestraightforwardly deduced from its component parts. Ex. (40) is an elliptic caseof this expression-type in that it suppresses kai; (41), on the other hand, exemplifies the full form, featuring kai.
33
Now the expression-type ta te alla kai X is in itself not relevant in the present study, simply because per does not occur in it. However, when the ex pression-type is put in a conditional form per does occur, modifying the conditional clause as a whole, while kai introduces the main clause: eiper tis alios,kai X 3 4 This expression-type is the idiomatic way (in post-Homeric Greek) of
saying 'especially X' in a conditional form: if anything, thenX. One example:(42) tên aretên phêis didakton einai, kai egô eiper allôi tôi anthrôpôn
peithoimên an, kai soi peithomai."You claim that virtue is something learnable, and if there is anyoneI beUeve on this point, it is you." (Plat. Prot. 329 B9).
It seems to me that this example can be placed in the same diachronicframework as (40)-(41). The fact that per in (42) is linguistically much moresalient (and hence predictable) than in (40)-(41) is, in the present connection,
of secondary importance.The decision to discuss (40)-(41) in connection with the later use ofper insubordinate clauses is supported by the fact that the boundary between whatmay be called the peripheral use oîper and the use of per in subordinate clausesis not sharp. Sometimes it does not make much difference whether we take agiven instance of per which occurs in a subclause as belonging to a single constituent or as belonging to the whole clause; in the first case we conceive ofperas comparable to (35)-(39), and in the second case we treat it as an instance ofsubclause- per in which per is separated from the subordinator. Consider:
(43) (Meriones to Deiphobus, when he has killed three Trojans - asagainst one Greek killed by Deiphobus:)
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 274/318
264 LINGUISTICS AND FORMULAS IN HOMER
Dêïphob'ê ara dê ti eïskomen aksion einai,treis henos anti pephasthai? epei su per eukheai houtô."Deiphobus! Do we judge rightly that three against one is a sufficientrecompense? When you are boasting that way..." (77.13,447).
A main clause for the epei-clause has to be supplied ('then I do the same').In this example it is possible to conceive of per as exclusively modifying su. Butit is important to realize that there is not much difference with the situation inwhich per has to be described as the subclause particle which it is to become inlater times; in the following example per is simply separated from its subordinator and no special emphasis is involved:
(44) (Melanthius the goatherd is abusing Odysseus, who is in disguise:) pantôs ouketi nôïdiakrineesthai oïô prin kheirôn geusasthai, epei su per ou kata kosmonaitizeis; eisin de kai allai daites Akhaiôn."I fancy that you and I will have to sample each other's fists beforewe say goodbye. For I don't like your way of begging. And anyhow
this house is not the only one where people dine." (Od.
20,181).
38
Other seemingly problematic instances, in which it is difficult to assign pera recognizable meaning, likewise, are actually not instances of Homeric per.Just as in (44), the particle has to be taken together with the subordinator ofthe subclause. Two examples:
(45) (Right through the morning the forces of Greeks and Trojans werein balance:)ernos de drutomos per anêr hôplissato deipnonoureos en bêssêisi (....)
têmos sphêi aretêi Danaoi rhêksanto phalangas.."But about the time when a woodman felling the tall trees in amountain dell prepares himself a meal, the Danaans put forth theirstrength and broke the enemy battalions." (Il. 11,86).
(46) (Here comes to the top of Mount Ida, where Zeus is watching thefighting:)hos d'iden, hos min erôs pukinas phrenas amphikalupsen,hoion hote proton per emisgesthên philotêti,es eunên phoitônte, philous lêthonte tokêas.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 275/318
PERIPHERAL INSTANCES 265"Zeus the Cloud-compeller saw her and at the first look his heartwas captured by desire, as in the days when they had for the firsttime enjoyed each other's love and gone to bed together withouttheir parent's knowledge." (77.14,295).
These are actually instances of êmo per and hote per, respectively.39 That is,they belong to the non-scalar use of per in subordinate clauses and not toHomeric per. Accordingly, they do not properly belong in the present chapter,in spite of the fact that they are, qua instances of 'subordinator + per',
peripheral too, just as the instances discussed in the present chapter. But todetermine with respect to which expression-type exx. (45)-(46) are peripheralis the subject of another study.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 276/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 277/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 278/318
268 LINGUISTICS AND FORMULAS IN HOMER14) Ex. (10) has the form of an original scalar expression-type, but as to its meaning it
points forwards to a later use of per. See further below (ex. (40)-(41) in 7.6) on other instances
of ' allos + per' in Homer.
15) Although alios per is non-scalar from a strictly semantic, context-independent point
of view, we may say that its use in the present example is nevertheless sufficiently justified by its
belonging to a directive speech act. The example is scalar as to the intention of the speaker: 'all
others do it; I want you to do it too', 'all others' being functionally equivalent to a scalar expres
sion.
16) Perhaps the linguistic justification for the poet to use per in the utterly non-scalar com
parative expressions (13) and (14) is the use of per in 'the same ay-contexts' (see 2.3.2.2, as well
as exx. (15)-(16) of ch. 1). This expression-type yields a number of bucolic clausulas, which may
have served as an analogon to (13)-(14), e.g. hos per epherte (Il. 2, 318), hoisi per arkhei ( Il . 2,
805), hoi per emeio (Il. 4, 324) etc. These instances of per do not belong to what I delimited in
1.1 as 'Homeric' per, they are not discussed in the present study.
17) As a formula (T2-expression), hothen te per oud' oiônoi is similar to hothen te per oinok-
hoeuei (Od. 21,142 = ex. (39) of ch. 4). See for the two instances of te per Ruijgh (1971: 480-
81).
18) This movement of per reflects the transition from a focus particle with a single term
as focus constituent to a particle which modifies a whole clause and which is placed, as an enclitic,
in the second place of the clause. The development in ' the same ay-contexts' (see 2.3.2.2) is
similar: *phroneeis ha egô per ⇒ phroneeis ha per egô. See also ex. (34) below.
19) See also Ruijgh (1971: 391-92). In attributing the concessive sense to kai in the rela
tive clause, Ruijgh attaches more importance to this particle than I would be inclined to do. In
my opinion, kai in the relative clause is a less than prototypical instance whose presence may
well be accounted for in terms of the third factor in 7.1.2 above: the solution of ad hoc metrical
problems.
20) Notice that hou te per oksutaton or hou per V oksutaton would have been metrically
possible. But in the prototypical instances of the expression-type in question te has no proper
place (see also the following note).
21) As to the occurrence of te in the relative clause, we may remark that when the removal
of per has no overt linguistic grounds, the introduction of te need not have them either. Notice,
however, that the use of te in (18) is not incompatible with the original generic meaning of theparticle. But this is not a particularly useful thing to say, because the relevant scalar property of
an eternal entity like the sun is by definition a 'permanent fact'.
22) Compare the similar formal development of 'kai + participle + per' to kaiper +
participle'.
23) Notably 'oudeper + participle'. The instances are collected in Oguse (1968).
24) To my mind, this does not only apply to the linguistic researcher, but to the native
speaker as well.
25) It follows that per should not be taken as modifying the whole clause so that it would
have to be discussed among the examples of ei per. For a different view see Denniston ( 1954:
488), who proposes 'even if really as a rendering of (28).
26) Notice that the concessive conditional (oud' ei..) is factual (see 6.2).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 279/318
NOTES TO CHAPTER 7 269
27) Comparable is Od. 8,187. Notice that these examples are listed by Denniston ( 21954:
482) under the heading 'Intensifying use', as instances of the original force of per. However, the
'force' of per in (38) is anything but old and original. See also ex. (46) below.
28) Presumably, the poet did not use ge in (35)-(39) because this particle is a too overtly
exclusive focus particle. And in the context of (35)-(39) the explicit exclusion of alternatives is
no more the point than the explicit inclusion of them. Instead what is called for is some vague,
not over-explicit sense of emphasis.
29) Of course, an easily definable diachronic status presupposes a well defined status in
the synchronic dimension. This is of course the problem with (35)-(39): their elusive diachronic
status is due to minimal synchronic prototypicality.
30) Compare Od. 17,273.
31) Compare Od. 23,7.
32) Per may perfectly be absent under semantically identical circumstances, see//. 23,483.
33) See Ruijgh (1971: 830-31). Here (40)-(41) are not discussed for the presence of per
but for the presence of te.
34) This use of ei per is, incidentally, entirely disconnected from any of the uses of ei per
discussed in ch. 6 above. No concession, scalarity or sceptic modality is involved.
35) For a full discussion of ei per tis alios within the framework of the use of ei per and hos
per in general in Attic Greek, see Bakker (in prep: ch. 6).
36) Ex. (42) is an instance of the use of per in focus contexts (see 2.3.2 above): the mainclause and the subclause have the same subject and predicate, which automatically lays emphasis
on the constituent by which they differ. This use of per is very frequent in later Greek, both in
relative and conditional clauses. It occurs already in Homer, but only in relative clauses; often
per is placed directly after the focused constituent. See ex. (34) of the present chapter as well as
ex. (15) of ch. 1 and ex. (10) of ch. 3. Full discussion of 'equative' per in focus contexts in Bak
ker (in prep).
37) See also Leaf (1902:35).
38) Compare Od. 2,327.
39) In Denniston (21954:482) per in (46) is assigned, wrongly, a very strong and 'original'
meaning: allegedly it intensifies prôton ('the very first time'). Again, we have the opposition be
tween 'old/original' and 'recent/peripheral' and again I opt for the second possibility. See also
ex. (38) above.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 280/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 281/318
8 OVERVIEW
The results and aims of the foregoing chapters can be summarized as follows. The central concern of the present study is the description of the use ofthe particle per in Homer. I have tried to present this description within a
perspective which is wider than the investigation of the use of a given linguistic item in a given Greek author. I have done so in the belief that the description of per benefits from it and in the hope that the present study has some valuefor other fields/disciplines than Greek linguistics proper. However, all the discussions in this study are primarily subservient to the description of per, onlyin the second place are they meant as contributions in their own right to thefields in question.
Conforming to its title, the present study has two major interests, (i) thelinguistic account of per and (ii) the use of per in formulas in particular and in
Homeric versification in general.Linguistics. The leading concept in the linguistic account of per is scalarity.
We speak of scalarity when a given constituent (word or phrase) represents thehighest, extreme value on a scale. A scale may either consist of properties/qualities (for example 'bravest' representing the high point of the 'brave-ness scale') or of persons/entities which are located on the scale by degree ofsome property/quality (for example, a scale with 'Achilles', the bravest warrior,as its high point). Scalarity may be encoded in language by means of scalar particles. Scalar particles are particles which mark a term that represents the high
point of a scale. In the analysis of scalar particles, this term may be called the focus constituent of the particle. Per in its most characteristic uses in HomericGreek is a scalar particle, just as kai in Greek and even in English.
It is intuitively clear that high points of scales have much to do with super-lativeness. What may be presented as highest on a scale has, by definition, somesuperlative quality. In English, superlativeness is a normal means to encodescalarity, in that a morphosyntactic superlative may convey a scalar sense by itself, without the aid of a scalar particle (for example, 'He could solve the mostdifficult problem'). The discussion of this use of the superlative (which is calledin the present study the scalar superlative) constitutes an important piece of the
background of the description of per.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 282/318
272 LINGUISTICS AND FORMULAS IN HOMER
Beside superlativeness, a second major concept in connection withscalarity is concession. Between a scalar expression and the sentence in whichit occurs, there is always an opposition. This is intuitively clear, too: the scalarexpression representing the extreme value on a scale expresses, by its very nature, the least likely or expected item in some situation ('Even Achilles (theleast likely person to do so) feared the battle'). When this fact is overtly ex
pressed, a concessive adverbial relation may arise: 'Even Achilles, who was {although he was) the bravest warrior, feared the battle'.
The description of per as a scalar particle hinges on the two concepts ofsuperlativeness and concession. Per is used when scalarity is expressed eitherin the form of a superlative or in the form of a concessive expression. In its mostcharacteristic use, per is used either in scalar superlative or in scalar concessivestatements. In the first case, per co-occurs with an adjective, being equivalentto the English scalar superlative, and in the second case it occurs in a participial
phrase or a non-restrictive relative clause.Adjectives, participles and relative clauses have one important property
in common. They are all terms with what may be called descriptive content: theyattribute a certain property to someone/something. As such they may be op
posed to terms without descriptive content: proper names and demonstrative pronouns. These terms only point ('refer') to a given entity, without attributing any property or quality to it. This distinction is fundamental for the presentstudy. It is dealt with in terms of the conceptual pair intension and extension.Constituents with descriptive content (adjectives etc.) are called intensionalterms, whereas constituents which lack descriptive content are called exten-sional.
In terms of the intension-extension distinction, per can be called the particle of intensional scalarity. This does not only appear from the fact that per in
its more characteristic uses co-occurs with intensional, and not with exten-sional, terms; scalar superlativeness and concession are in themselves intensional phenomena, because they always involve descriptive content. Thealternative scalar particle in Greek, kai, on the other hand, is the particle of ex-tensional scalarity; it may have an extensional term as its focus constituent, justas even in English.
Sometimes per and kai co-occur. This co-occurrence reflects the intimatesemantic bond between scalarity, superlativeness and concession. Kai marksthe extensional side of the scalarity, taking a term without descriptive contentas focus constituent (kai Akhilleus, kai hos - 'even Achilles', 'even so'). Per, on
the other hand, marks the intensional side of the scalarity: occurring in a phrase
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 283/318
OVERVIEW 273
which adds descriptive content, it explains why the focus constituent of kai functions as such. The result is a phrase which allows of a concessive interpretation:'Even Achilles, the bravest warrior...' ⇒ 'Achilles, even though he is the bravestwarrior'.
The relation between scalarity and concession may be conceived of assemantic in the sense that the above mentioned opposition of a scalar expression and the sentence in which it occurs springs from the very meaning of thescalar expression. However, in the case of per more is involved. The relation
here between scalarity and concession has a strong diachronic orientation, tothe effect that the semantic connection may easily turn into a transition in time.The scalar (or more precisely: superlative) force of per is diachronically unstable. After its disappearance, it leaves per as a marker of simple concession,especially in its use with participles, the combination 'participle + per' becoming the normal concessive adverbial constituent in Homeric Greek. This transition, which can be parallelled in other languages, is very natural, precisely
because scalarity and concession are intimately connected semantically.Scalar phenomena in language are most often discussed in (formal)
semantic terms, often without any attention for the way they are actually usedin discourse. In the present study I have tried to fill this gap. The most centraluse of scalar expressions, in any case that of per in Homer, seems to be the statement of what I call a superlative fact. A sentence in which a scalar expressionoccurs is often uttered not for its own sake; rather, it serves to assert somethingremarkable in an indirect way. For example, a sentence like 'Even Achillesfears the battle' is preferably uttered in discourse to assert that the battle inquestion is terrifying, not that it is feared by Achilles. On account of the scalar(superlative) force of 'Achilles', the sentence is an argument for the superlative fact that the battle was terrible. The specific use of scalar expressions in
discourse as arguments for superlative facts provides a basis for the differentiation of the scalar instances of per from the merely concessive ones.
Beside scalarity, there is a second important concept in the linguisticdescription of r' prototypicality. Prototypicality does not apply exclusively to
per; it is an important concept in descriptive linguistics in general. When working in the framework of prototypicality, we acknowledge the fact that not all instances of a given linguistic sign are equally salient and characteristic. Some aremore characteristic of their kind (prototypical, we say) than others.Prototypicality seems to be the right way to deal with the residue that is con
comitant with the description of any linguistic item, the instances that do not,in being exceptional and/or deviant, conform to the description. This residue
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 284/318
274 LINGUISTICS AND FORMULAS IN HOMERneed not be troublesome, as long as it is recognized that language (use) admitsa considerable amount of variability which may imply divergence from thenorm which is set by the prototypical instances.
In the case of per and its description, the prototype way of thinking may be appUed to gauge the degree to which instances conform to the properties ofintensional scalarity mentioned above. Some instances of per are definitelymore scalar than others, and some are not scalar at all. In this respect, super-lativeness appears to be an important prototypical property of per, whose
presence or absence determines to a high degree the prototypicality of a giveninstance.
Formulas. The second major interest of the present study may be seen asan interesting complicating factor for the prototype way of thinking. It isspecific for the Homeric context in which per occurs. A given instance of permay not be what it 'should' be as a prototypical instance, just because of the
peculiarities of the Homeric text. 'Homer' is a metrical text with many characteristics pointing to oral composition or at least to cognation with orally com
posed texts. These characteristics cannot but leave their imprint on the
linguistic form of the Iliad and Odyssey.For the Homeric poet, the all-important thing is to produce his verses asautomatically as possible and to avoid interruption in the flow of poetic production. To achieve this end, he may make use of what are called 'formulas' in theParryan tradition of Homeric criticism (and indeed in most oral poetry studies).Formulas are considered to be ready-made building-blocks, which facilitate the
poet's rapid production of hexameter lines. They are commonly thought to beformulaic by themselves, in the sense that they are different from ordinary language expressions, both as to their form and as to their meaning. In this ap
proach one is led to believe that, in being artificial and formulaic, the Homericlanguage is something qualitatively different from ordinary language.However, from a linguistic point of view I want to modify this view.
Instead of the structural view of the Homeric formula meant above, I endorse a functional view of Homeric formulas and versification. I treat formulasnot as a discrete phenomenon that has to be defined and explained, but as oneof a number of possible consequences of oral composition. My claim is that'formulas' should not be isolated and set off against ordinary language expressions, because in Homer the variability which is present anyway in the languageis fully exploited, so as to achieve a system of versification which enables the
poet to produce a constant flow of verses. Only rarely are the boundaries of linguistic possibilities transgressed, precisely because any linguistic sign or expres-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 285/318
OVERVIEW 275
sion has instances which range from prototypical to less prototypical('peripheral') . It is with the less prototypical instances that the linguistic studyof 'formulas' in Homer is mostly concerned.
For the study of the use of per in Homeric versification, two main 'sources'of lesser prototypicality may be distinguished. Each may be treated as a consequence of oral composition.
1) Semantic integration of formulas. A given instance of per, which is partof a recurrent phrase ('formula' in the Parryan sense of standardized building-
block) may be used, under formular pressure, in a semantic or syntactic environment for which it was not originally devised. The result is, in linguisticterms, a less prototypical instance of the expression in question.
2) Redundant semantics. Often, it is assumed that formulas have oneatomic meaning. More justice is done, however, to the facts of Homeric versification when one assumes that formulaic phrases may often be subdividedinto a nucleus, which constitutes the semantic identity of the phrase, and a
periphery, which is semantically redundant. This insight is especially useful inconnection with the metrical form of formulas. By adding or omitting thesemantically redundant material, the poet may deUberately alter the metricalform of phrases so as to make them fit into one of the metrical slots (cola) ofthe dactylic hexameter.
This phenomenon, which to my mind is crucial for Homeric versification,can be very well observed in the case of the use of per with participles as a concessive adverbial constituent. The nucleus of the expression is formed by thecombination 'participle + per', which may be localized in the verse in variousways, according to the metrical form of the participle. The semantically redundant material is added in the form of other particles (kai, mala, empês) by whichthe phrase is extended (forwards or backwards) to the nearest metrical bound
ary. Sometimes per itself is the redundant element which may be deliberatelyadded or omitted. Greek epic diction exploits here in a systematic way the
phenomenon of strengthening, which is very common in ordinary language: intheir less prototypical uses, some linguistic items, especially particles, may beused to strengthen another expression (compare even in even though). The sameapplies, in linguistic terms, to kai in 'kai + participle + per', but here thestrengthening element is there for the sake of easy and smooth versification.
The description of per offered in the present study does not cover all theinstances of the particle in Homer. Only those instances are dealt with that may
in some way or another be discussed in terms of concession and scalarity. Theseinstances may be rightly called 'Homeric' per, since after Homer scalar-conces-
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 286/318
276 LINGUISTICS AND FORMULAS IN HOMERsive per has only a marginal existence in the language. In most instances by farin post-Homeric Greek, per is used as an enclitic adhering to subordinatingconjunctions. This use occurs as early as Homer, but it yields precisely thoseinstances that have been left out of account in the present study.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 287/318
APPENDIX
This Appendix Hsts all the instances of per in Homer that can be found inGehring (1891-95). Beside the instances of 'Homeric' per, which are the sub
ject of the present study, I have included the instances of post-Homeric per (notdiscussed). The former are specified according to the expression-type underwhich they are discussed. The latter are left unspecified; they have been merely included to complete the statistical picture. Sometimes exact classificationis difficult, for example in the case of the non-concessive instances of ei per,which may equally be Hsted under Homeric and post-Homeric per. Within the
class of Homeric per , too, there are instances which defy unequivocal classification. But within the framework of the present study (see 1.4), this is a readilyaccepted phenomenon. To avoid unnecessary problems in classification, I havelisted the peripheral instances of chapter 7 in one single block. However, thedivergent instances that belong as to their form to one of the regular expression-types (see 7.3.1) are Hsted under the original expression-type. Finally, tex-tuaHy doubtful instances are given in square brackets. After the Hsting of theinstances, the total figures are presented in a separate table.
1) Scalar superlatives ('adjective + per'), see 3.1.
Eiad: 4,421; 9,514,554; 13,300; 14,217; 16, 638; 18,108; 20,65; 21, 63.Odyssey: 3,321; 5,73; 14,464.
2) Scalar-concessive relative clauses ('kai/oude X, hos per'), see 3.2. Eiad: 6, 100; 7, 114; 9, 498; 14, 246, 344; 16, 709; 18, 118; 19, 95, 416; 20,
358;21,107,196; 24, 603.Odyssey: 7,322; 13,249; 20,46.
3) Scalarity and the comparative (ê per), see 3.3. Riad: 1,260; 8,190; 10,556; 11,319; 14,468; 16,688; 18,302; 24,504.
Odyssey:4,819;8,154;17,417.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 288/318
278 LINGUISTICS AND FORMULAS IN HOMER
4) Negative polar scalarity, see 3.4. Iliad: 11,391; 14,416.Odyssey: 8,212,547; 10,24.
5) Scalarity in wishes and commands, see 3.5. Iliad: 1,353,508; 8,242,243; 9,301; 11,796; 12,349,356, 362; 14,358; 16,
38,205; 17,121, 634,712; 19, [189]; 20,119,300; 21,308; 23,97.Odyssey: 4,35; 6,325; 18,122; 20,199.
6) Unclassifiable scalar instances, see 3.6. Ilid: 8,353,452; 9,110; 13,415; 15,588; 17,239; 22,73; 23,79; 24,749.Odyssey: 1,236; 4, 97; 11,441,501; 19,581; 21,79.
7) Scalar-concessive participles following kailoud' hôs, see 4.2. Riad: 3,159; 11,721,841; 15,617.Odyssey: 1,6; 5,324; 11,88,104.
8) Scalar-concessive participles, see 4.3.1. Iliad: 1, 217,241, 275,586, 587, 588; 2,246,270; 3,201; 4, 534; 5, 94, 382,
571,625; 6,85,360; 7,110; 8,125,253,317; 9,198,518,552; 10,114,549; 11, 418,554; 12,178,410; 13,57,142, 317,419, 630; 14,33, 58,128,260,375,379; 15,164,185,195,399,450,585,604,651; 16, 620,624; 17,181,276,292,459,571,663,710; 18,112,126,273; 19,8, 65,80, 82, 142, 155,189; 20, 356; 21, 384; 22, 416, 424; 23, 306; 24, 53,104,298,523.
Odyssey: 1,288,309; 2,200,219,249; 3,240,284; 4,104,416,502,549,553,733; 5, 209, 341; 6, 87; 7, 215, 297; 8, 316, 360, 478; 9, 379; 10, 174,189,246; 11,265,350; 12,271,340; 13,280; 14,142,155,298; 15,49,361; 16,147; 17,555,570; 18,165,178,385; 19,324,511; 20,131,271,
274; 21,103,250; 22,172; 23,12,82,230,361; 24,499.
9) Concessive participles, specified:
9a) Non-superlative participles, see 4.3.2.mad: 1,546,577; 4,387,5,135; 8,99,284; 9,343,605,627; 10,448; 13,361;
14,1, 98; 15,30,407,476; 16,550,617, 815; 17,229,539; 18,549; 20,
21, 436; 21, 185,483; 22, 218; 23, 610; 24,20, 35, 423,428, 570,593,609, 617,750.
Odyssey: 6,136; 7,224; 8,331; 9,57; 10,441; 11, 111, 425; 12,138; 14,310;
16,264; 18,21; 19,253,356; 21,370; 23,7.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 289/318
APPENDIX 279
9b) Dependent participles, see 4.3.3. Iliad: 1,352.Odyssey: No instances.
9c) Modally embedded participles, see 4.3.4. Iliad:l, 131; 9,373; 15,133.Odyssey: 1, 315; 4,284; 16,430; 17,13,47; 21,129; 22,409.
10) Conditionals (eiper), specified:
10a) Concessive, see 6.3.1. Iliad: 1, 81; 2, 597; 3,25; 4, 55,160; 5,224; 7,117; 8,153; 10,115,225; 11,
116; 12, 223, 245, 302; 15, 99, 117; 16, 847; 19, 32, 164; 21, 576; 22,191,487.
Odyssey: 1,167, 204; 2,246; 5,40; 7,321; 11,113; 12,140; 13,138,143; 16,276; 17,14; 18,318; 20,49.
10b) Non-concessive, see 6.4.3. Iliad: 1,580; 2,123; 4,261; 5,232; 7,387; 8,205; 13,288,464; 16,263; 22,
86; 24, 667.Odyssey: 1,188; 8,355,408; 9,35; 20,42.
11) Peripheral instances (unspecified), see ch. 7. Iliad: 1, 416; 2, 236; 7,204; 8, 201; 10,70; 11, 86, 453,789; 12, 361; 13, 72,
447, 573; 14,125,295; 15, 553; 16,31,245, 523; 17,104; 18,151; 19,57,200,217; 20,100; 21,130,410,441; 22,389; 24,130,235.
Odyssey: 1,59; 3,236; 4,214,376,379,468,729; 5,29; 6,282; 8,187; 10,551;11, 452; 13, 294; 14, 392; 15, 540; 17, 273; 19, 541; 20, 181; 22, 473;23,209.
Other instances (this is an unspecified enumeration of the instances of 'post-Homeric' per (see 1.1):
Iliad : 1,211; 2,258,286,293,318,805, 861; 3,3; 4,236,259,263,324,361,524; 5, 265,340,477, 802, 806; 6,41,146,292, 379,384, 398,477; 7,159,286,370; 8,415; 9,46,310,367; 10,7,309,396,568; 11,126; 12,33,118,256,346,359; 13,101,524,638; 14,50,131,319,323; 15,225,675,707; 16,557,618; 17,232,509, [587], 720; 18,263,293,363,422,518; 19,42; 20,123,188; 21,4,55,554; 22,250; 23,84,480,583, 659,782, 802; 24,201,348,382,398,487.
Odyssey: 1,128,210; 2,56,305,327,333; 4,565,627; 5,82,188; 7,55,203,206,312; 8,31,107,510; 9,365,543; 10,103,240,279,285,435,483;
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 290/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 291/318
REFERENCES
Adkins, A.W.H. 1960. Merit and responsability, Oxford: O.U.P.Altmann, Η . 1976. Die Gradpartikeln im Deutschen. Untersuchungen zu ihrer
Syntax, Semantik und Pragmatik, Tübingen: Niemeyer.Austin, N. 1975. Archery at the dark of the moon. Poetic problems in Homer's
Odyssey, Berkeley, Los Angeles, London: Univ. of Calif. Press.Auwera, J. van der 1983. 'Conditionals and antecedent possibilities', Journal of
Pragmatics 7, 297-309. — . 1984. 'Maar en alleen als graadpartikels', in Van der Auwera & Van-
deweghe, eds., 105-18. — . 1985. Language and logic. A speculative and condition-theoretic study,
Amsterdam: Benjamins (Pragmatics & Beyond).Auwera, J. van der, ed. 1980. The semantics of determiners, London, Baltimore:
Croom Helm.Auwera, J. van der; Vandeweghe, W., eds. 1984. Studies over Nederlandse par
tikels, Wilrijk: Universiteit van Antwerpen.Baker, G.P.; Hacker, P.M.S. 1984. Language, sense and nonsense, Oxford:
Blackwell.Bakker, EJ. 1986. 'Hosper en eiper : een aspect van Attische conversatie', Lam
pas 19,142-158. — . 1988a. 'Restrictive conditionals', in Rijksbaron, ed. (1988), — . 1988b. 'Long diphthongs and hiatus in early Greek epic: phonology and
the role of epic diction', Mnemosyne 40, — . in prep. Identification and equation. A study of relative and conditional
clauses in Classical Greek, (working title).Bakker, W.F. 1966. The Greek imperative. An investigation into the aspectual dif-
ferences between the present and aorist imperatives in Greek prayer from Homer up to the present day, Amsterdam: Hakkert.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 292/318
282 LINGUISTICS AND FORMULAS IN HOMER
Bal, M. 1977. Narratologie, Paris.Beekes, R.S.P. 1972. 'On the structure of the Greek hexameter. O'Neill inter
preted.', Glotta 50,1-10.Bennett, J. 1982. 'Even if, Linguistics and Philosophy 5,403-18.Benveniste, E. (1948) 1975. Noms d'agent et noms d'action en Indo-Européen,
Paris: Librairie d'Amérique et d'Orient.Bierwisch, M.; Heidolph, K.E., eds. 1968. Progress in linguistics, The Hague:
Mouton
Bolkestein, A.M. 1980. Problems in the description of modal verbs. An investigation of Latin, Assen: Van Gorcum. — . 1987. 'Discourse functions of predications: The background/foreground
distinction and tense and voice in Latin main and subordinate clauses', in Nuyts & De Schutter, eds. (1987).
Booth, W.C. 1961. The rhetoric of fiction, Chicago: The University of ChicagoPress.
Brugmann, K. - Thumb, A. 41913. Griechische Grammatik. Lautlehre, Stamm-bildungs- und flexionslehre, Syntax, München: Beck.
Buck, CD. 1955. The Greek dialects, Chicago: The University of Chicago Press.
Burton-Roberts, N. 1975. 'Nominal apposition', Foundations of Language 13,391-419.
Carlson, G. 1982. 'Distribution of free choice any' Chicago Linguistic Society17, 8-23.
Chantraine, P. 1948. Introduction à Vüiade, Paris: Les belles lettres. — . 1953. Grammaire homérique, Vol. 2 Syntaxe, Paris: Klincksieck. — . 1968 ff. Dictionnaire étymologique de la langue grecque, Paris: Klincksieck.Cole, P., ed. 1978. Pragmatics ( = Syntax and semantics vol. 9), New York, San
Francisco, London: Academic Press.
Cole, P.; Morgan, J.L., eds. 1975. Speech acts ( = Syntax and semanticsvol. 3), New York, San Francisco, London: Academic Press.Combellack, F.M. 1965. 'Some formulary illogicalities in Homer', Transactions
of the American Philological Association 96,41-56.Cruse, D.A. 1986. Lexical semantics, Cambridge: C.U.P. (Cambridge
Textbooks in Linguistics).Davison, A. 1980. 'Any as universal or existential?', in Van der Auwera, ed.
(1980), 11-40.Denniston, J.D. 21954. The Greek particles, Oxford: O.U.P.Dik, S.C. 1978. Functional grammar, Dordrecht: Foris.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 293/318
REFERENCES 283
Donellan, . 1966. Reference and definite descriptions', Philosophical Review75,281-304 (Reprinted in Steinberg & Jakobovits, eds. (1971), 100-114.
Ducrot, O.1972. Dire et ne pas dire, Paris: Hermann. — . 1980. Les Échelles argumentatives, Paris: Minuit.Ebeling, H., ed. 1880-85 Lexicon Homericum, Leipzig: Teubner.Edwards, M.W. 1987. Homer. Poet of the Iliad, Baltimore/London: John Hop
kins University Press.Eikmeyer, Η -J., Rieser, H., eds. 1981. Words, worlds and contexts. New ap
proaches in word semantics, Berlin: de Gruyter.Fauconnier, G. 1975a. 'Pragmatic scales and logical structure', Linguistic In
quiry 6,353-375. — . 1975b. 'Polarity and the Scale Principle' Chicago Linguistic Society 11,188-
99.
— . 1979. 'Implication reversal in natural language', in Guenthner & Schmidt,eds., 289-301.
Fillmore, C.J. 1975. 'An alternative to checklist theories of meaning', Proceed
ings of the 1st annual meeting. Berkeley Linguistic Society, 123-31.
Fillmore, C.J.; Langendoen, D.T., eds. 1971. Studies in linguistic semantics, NewYork: Holt.
Fleischman, S. 1982. The future in thought and language. Diachronic evidence
from Romance, Cambridge: C.U.P. (Cambridge Studies in Linguistics).
Ford, CE.; Thompson, S.A. 1986. 'Conditionals in discourse. A text-basedstudy from EngUsh', in Traugott et al., eds. (1986), 353-372.
Fox, . 1983. 'The discourse function of the participle in Ancient Greek', inKlein-Andrieu, ed. (1983), 22-41.
Fraenkel, H. 1925. 'Griechische Wörter', Gioita 14,1-13. — . 1926. 'Der kalHmachische und homerische Hexameter ', Nachrichten der
Gesellschaft der Wissenschaften zu Göttingen (Phil.-Hist. Klasse), 197-229.Fraser, . 1971. 'An analysis of Even in English', in Fillmore & Langendoen,
eds. (1971), 151-180.Frege, G. 1892. 'Über Sinn und Bedeutung', Zeitschrift f. Philosophie und
Philosoph. Kritik 100,25-50. Reprinted in Patzig, ed. (1975), 40-65.Frisk, H. 1960-72. Griechisches etymologisches Wörterbuch, Heidelberg: Carl
Winter.Gazdar, G. 1979. Pragmatics: implicature, presupposition and logical form, New
York: Academic Press.Gehring, A. [1891-95] 1970. Index Homericus, Leipzig (reprint Hildesheim:
Olms).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 294/318
284 LINGUISTICS AND FORMULAS IN HOMER
Genette, G. 1972. Figures III, Paris: Seuil.Givón, T. 1970. 'Notes on the semantic structure of English adjectives', Lan-
guage 46. — . 1973a. 'Opacity and reference in language: an inquiry into the role of
modalities', in Kimball, ed. (1973), — . 1973b. 'The time-axis phenomenon', Language 49, 890-925 — . 1979. On understanding grammar, New York, San Francisco, London:
Academic Press.
— . 1982. 'Logic vs. pragmatics, with human language as the referee: towardsan empirically viable epistemology', Journal of Pragmatics 6, 81-133. — . 1984. Syntax, a functional-typological introduction,vol. 1, Amsterdam: Ben
jamins. — . 1985. 'Iconicity, isomorphism and non-arbitrary coding in syntax', in
Haiman, ed. (1985), 187-219.Givón, T., ed. 1979. Discourse and syntax ( = Syntax and semantics, vol. 12),
New York, London, Toronto: Academic Press. — . 1983. Topic continuity in discourse: a quantitative cross-language study,
Amsterdam/Philadelphia: Benjamins (Typological Studies in Language3).
Goodwin, W.W. 1889. Syntax of the moods and tenses of the Greek verb, London: Macmillan.
Greenberg, J.H., Ferguson, CA., Moravcsik, E.A., eds. 1978. Universals· ofhuman language, vol. 4: Syntax, Stanford, Calif: Stanford University Press.
Grice, H.P. 1975. 'Logic and conversation', in Cole & Morgan, eds. (1975), 41-58.
Guenthner, F.; Schmidt, SJ. 1979. Formal semantics andpragmatics for naturallanguages, Dordrecht: Reidel.
Haack, S. 1978. Philosophy of logics, Cambridge: C.U.P.Haas, W. 1962. 'The theory of translation', Philosophy 37,208-28.Haiman, J. 1974. 'Concessives, conditionals, and verbs of volition', Foundations
of Language 11,341-59. — . 1978. 'Conditionals are topics', Language 54,564-89. — . 1980. 'The iconicity of grammar: isomorphism and motivation', Language
56,514-40.Haiman, J., ed. 1985. Iconicity in syntax, Amsterdam, Philadelphia: Benjamins
(Typological Studies in Language 6).Hainsworth, J.B. 1964. 'Structure and content in epic formulae: The question
of the unique expression', Classical Quarterly n.s. 14,2,155-64.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 295/318
REFERENCES 285
— . 1968. The flexibility of the Homeric formula, Oxford: O.U.P.Haudry, J. 1973. Tarataxe, hypotaxe et corrélation dans la phrase latine', Bul
letin de la Société de Linguistique de Paris 68 (1), 147-86.Hawkins, JA. 1978. Definiteness and indefiniteness. A study in reference and
grammaticality prediction, London, New York: Croom Helm.Hoeksema, J. 1983. 'Negative polarity and the comparative', Natural Language
and Linguistic Theory 1,403-434. — . 1984. To be continued: the story of the comparative', Journal of Seman-
tics 3,93-107.Hoekstra, A. 1965. Homeric modifications of formulaic prototypes. Studies in the
development of Greek epic diction, Amsterdam, London: North-Holland. — . 1981 Epic verse before Homer. Three studies, Amsterdam, Oxford, New
York: North-Holland.Hoffmann, M.E. 1987. Negatio Contrarii A study of Latin litotes, Assen,
Maastricht: Van Gorcum (Studies in Greek and Latin Linguistics).Hopper, P.J. 1979. 'Aspect and foregrounding in discourse', in Givón, ed.
(1979),213-241.Hopper, P.J., ed. 1982. Tense-aspect: Between semantics and pragmatics,
Amsterdam/Philadelphia: Benjamins (Typological Studies in Language
1).Hopper, P.J.; Thompson, S.A 1980. Transitivity in grammar and discourse',
Language 56,251-99. — . 1985. 'The iconicity of the universal categories "noun" and "verb"', in
Haiman, ed. (1985), 151-83.Horn, L.R. 1969. 'A presuppositional analysis of only and even', Chicago Lin
guistic Society 5, 98-107. — . 1972. On the semantic properties of logical operators in English, UCLA diss.
(Reprinted by Indiana University Linguistics Club (1976)). — . 1978a. 'Some aspects of negation', in Greenberg et al., eds. (1978), 127-
210. — . 1978b. 'Remarks on neg-raising', in Cole, ed. (1978), 129-220.Ickler, N.L. 1977. Topicalization and relativization in Old Russian', Berkeley
Linguistic Society 3, 656-69. — . 1981. The particle 'ze' in Old Russian: the discourse origins of conditionals
and relatives, Diss. Berkeley, Calif.Ingalls, W.B. 1970. 'The structure of the Homeric hexameter. A review',
Phoena 24,1-12.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 296/318
286 LINGUISTICS AND FORMULAS IN HOMER
— . 1976. 'The analogical formula in Homer', Transactions of the AmericanPhilological Society 106,211-26.
— . 1979. 'Formular density in the similes of the Iliad, Transactions of the American Philological Association 109, 87-109.
Jacobs, J. 1983. Fokus und Skalen: Zur Syntax und Semantik der Gradpartikelnim Deutschen, Tübingen: Niemeyer.
Janko, R. 1982. Homer, Hesiod and the Hymns. Diachronie development in epicdiction, Cambridge: C.U.P. (Cambridge Classical Studies).
Jones, Α .; Churchhouse, R.F. eds. 1976. The computer in literary and linguisticstudies: proceedings of the third international symposium, Cardiff: University of Wales Press.
Jong, I.J.F. de 1987. Narrators and focalizers. The presentation of the story in the
Iliad, Amsterdam: Grüner.Karttunen, L.; Peters, S. 1979. 'Conventional implicature' in Oh & Dinneen,
eds. (1979), 1-56.Kempson, R.M. 1975. Presupposition and the delimitation of semantics,
Cambridge: C.U.P (Cambridge Studies in Linguistics 15)Kimball, J.,ed. 1973. Syntax and semantics, vol. 2, New York/London: Seminar
Press.Kiparsky, P. 1976. Oral poetry: some linguistic and typological considerations',
in Stolz & Shannon, eds. (1976), 73-106.Kiparsky, P.; Kiparsky, 1968. 'Fact', in Bierwisch & Heidolph, eds. (1968),
143-173. Reprinted in Steinberg & Jakobovits, eds. (1971), 345-369.Kirk, G.S. 1966a. 'The structure of the Homeric hexameter', Yale Classical
Studies 20, 76-104. — . 1966b. Tormular language and oral quality', Yale Class. St. 20, 153-74.
Reprinted in Kirk (1976).
-—. 1976. Homer and the oral tradition, Cambridge: C.U.P. — . 1985. The Iliad: a commentary, Vol I: books 1-4, Cambridge, C.U.P.Kirsner, R.S. 1985. 'Iconicity and grammatical meaning', in Haiman, ed. (1985),
249-270.Klein-Andrieu, F., ed. 1983. Discourse perspectives on syntax, New York, Lon
don etc: Academic Press.Kneale, W; Kneale, M. 1962. The development of logic, Oxford: UP.König, E. 1981. 'The meaning of scalar particles in German', in Eikmeyer,
Rieser, eds. (1981), 107-132. — . 1986. 'Conditionals, concessive conditionals and concessives: areas of con
trast, overlap and neutralization', in Traugott et al, eds. (19861229-46
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 297/318
REFERENCES 287
Kühner, R. - Gerth, B. 1898-1904. Ausführliche Grammatik der griechischen
Sprache Π : Satzlehre. Hannover.Ladusaw, W. 1979. Polarity sensitivity as inherent scope relations. Ph.D. diss.
University of Texas at Austin.Latacz, J., ed. 1979. Homer. Tradition und Neuerung, Darmstadt: Wis
senschaftliche Buchgesellschaft.Leaf, W. 1900-2. The Iliad, edited with apparatus criticus, prolegomena, notes
and appendices, London (repr. 1971 Amsterdam: Hakkert).
Leech, G.N. 1983. Principles of pragmatics, London, New York: Longman.Lehmann, Chr. 1984. Der Relativsatz, Tübingen: Narr.Levinson, S.C. 1983. Pragmatics, Cambridge: C.U.P (Cambridge Textbooks in
Linguistics),liddell , H.G. & Scott, R. 919404 Greek-English lexicon, Oxford: O.U.P.Linsky, L., ed. 1971. Reference and modality, Oxford: O.U.P. (Oxford Readings
in Philosophy).Lord, A.B. 1960. The singer of tales, Cambridge, Mass; London: Harvard
University Press.Lyons, J. 1977. Semantics 2 vols., Cambridge: C.U.P.
Matthews, P.H. 1981. Syntax, Cambridge: C.U.P. (Cambridge Textbooks inLinguistics).
Meister, K. 1921. Die homerische Kunstsprache, Leipzig: Teubner.Mill, J.S. 1843 A system of logic, London: LongmansMinton, W.W. 1965. The fallacy of the structural formula', Transactions of the
American Philological Association 96,241-53.Momo, D.B. 1891. A grammar of the Homeric dialect, Oxford: UP.Monteil, P. 1963. La phrase relative en grec ancien, Paris: Klincksieck.Murray, G. 41934. The rise of Greek epic, Oxford: O.U.P.
Nagler, M.N. 1967. Towards a generative view of the oral formula,
) Transac-tions of the American Philological Association 98,269-311. Notopoulos, J.A. 1962. The Homeric Hymns as oral poetry', American Jour-
nal of Philology 83,337-68. Nuyts, J.; De Schutter, G., eds. 1987. Principles of Word Ordering, Dordrecht:
Foris.Oguse, A. 1962. Recherches sur le participe circonstanciel en grec ancien, Paris:
Klincksieck. — . 1968. 'Sur les concessives à principale négative', Revue de Philologie XLII,
262-84.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 298/318
288 LINGUISTICS AND FORMULAS IN HOMEROh, C.-K.; Dinneen, D.A., eds. 1979. Presupposition (= Syntax and semantics
vol. 11), New York, San Francisco, London: Academic Press.O'Neill Jr., E.G. 1942. 'The localization of metrical word-types in the Greek
hexameter. Homer, Hesiod and the Alexandrians', Yale Classical Studies,8,105-78.
Paardekoper, P.C. 1979. Ook maar iemand', De Nieuwe Taalgids 72,429-448.Packard, D.W. 1976. 'Metrical and grammatical patterns in the Greek
hexameter', in Jones & Churchhouse, eds. (1976), 85-91.
Palmer, F.R. 1986. Mood and modality, Cambridge: C.U.P. (CambridgeTextbooks in Linguistics).
Parry, M. 1928a. L'Épithète traditionelle dans Homère; Essay sur un problèmede style homérique, Paris: Société éditrice des Belles Lettres.
— . 1928b. Les formules et la métrique d'Homère, Paris: Société des Belles Lettres.
— . 1930. 'Studies in the epic technique of oral verse-making. I. Homer andHomeric style', Harvard Studies in Classical Philology 41,73-147.
— . 1932. 'Studies in the epic technique of oral verse-making. Π . The Homericlanguage as the language of an oral poetry', Harvard Studies in Classical
Philology 43,1-50.-—. 1971. The making of Homeric verse. The collected papers of Milman Parry,
edited by Adam Parry, Oxford: O.U.P.Patzig. G., ed. 1975. Gottlob Frege. Funktion, Begriff, Bedeutung. Fünf logische
Studien, Göttingen: Vandenhoeck & Ruprecht.Peabody, B. 1975. The winged word, Albany: State Univ. of New York Press.Pinkster, H. 1972. On Latin adverbs, Amsterdam: North-Holland.Porter, H.N. 1951. "The early Greek hexameter', Yale Classical Studies 12, 1-
63.
Prendergast, G.L. 1962. Complete concordance to the Iliad of Homer. New edition, revised and enlarged by Benedetto MarzuUo.Quine, W.V.0.1953. Reference and modality', in From a logical point of view,
Cambridge, Mass: Harvard Univ. Press. Reprinted in Linsky, ed. (1971),17-34.
Quirk, R.; Greenbaum, S.; Leech, G.; Svartvik, J. 1972. A grammar of contem porary English, London: Longman.
Raalte, M. van 1986. Rhythm and metre. Towards a systematic description ofGreek sty chic verse, Assen: Van Gorcum (Studies in Greek and Latin Linguistics).
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 299/318
REFERENCES 289
Ramsey, E. 1987. 'The functional distribution of preposed and postposed "if"and "when" clauses in written discourse', in Tomlin, ed. (1987), 383-408.
Rescher, Ν . 1968. Topics in philosophical logic, Dordrecht: Reidel.Rieu, E.V. 1946. Homer, the Odyssey (transi.), Harmondsworth: Penguin. — . 1950. Homer, the Iliad (transl.), Harmondsworth: Penguin.Rijksbaron, A. 1984. The syntax and semantics of the verb in Classical Greek. An
introduction, Amsterdam: Gieben — . 1986. 'The pragmatics and semantics of conditional and temporal clauses.
Some evidence from Dutch and Classical Greek', Working Papers in Functional Grammar 13, Amsterdam.
Rijksbaron, Α ., ed. 1988. In the footsteps of Raphael Kühner, Amsterdam:Gieben.
Rosch, E.H. 1973. 'Natural categories', Cognitive Psychology 4,328-50. — . 1978. 'Principles of categorization', in Rosch & Lloyd, eds. (1978), 27-48.Rosch, E.H.; Lloyd, ., eds. 1978. Cognition and categorization, Hillsdale, N.J.:
Erlbaum Associates.Ruijgh, . J. 1957. L'élément achéen dans la langue épique, Assen: Van Gorcum. —.1971.Autour de 'te épique'. Études sur la syntaxe grecque, Amsterdam: Hak-
kert. — . 1985. 'L'emploi 'inceptif du thème du présent du verbe grec', Mnemosyne
38,1-61.Russell, . 1905. 'On denoting', Mind 14,479-93.Russo, J.A. 1963. 'A closer look at Homeric formulas', Transactions of the
American Philological Association 94,235-47. — . 1966. 'The structural formula in Homeric verse', Yale Classical Studies 20,
219-40. — . 1976. 'Is "oral" or "aural" composition the cause of Homer's formulaic
style?', in Stolz & Shannon, eds. (1976), 31-54.Saussure, F. de (1916) 1982. Cours de linguistique générale, édition critique
préparée par Tullio de Mauro, Paris: Payot.Schmerling, S. 1971. 'A note on negative polarity', Papers in Linguistics 4,1.Schmidt, CE. 1885. Parallel-Homer oder Index aller homerischen Iterati in lex
ikalischer Anordnung, Göttingen: Vandenhoeck & Ruprecht.Schoorl, S. 1980. 'Opacity and transparency: a pragmatic view', in Van der
Auwera, ed. (1980), 156-165.Schwyzer, E. - Debrunner, A 1950. Griechische Grammatik, 2. Band: Syntax
und syntaktische Stilistik, München: Beck.Searle, J. 1975. 'Indirect speech acts', in Cole & Morgan, eds. (1975), 59-82.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 300/318
290 LINGUISTICS AND FORMULAS IN HOMER
Seuren, P.A.M. 1985. Discourse semantics, Oxford: Blackwell.Shipp, G.P. 21972 Studies in the language of Homer, Cambridge: U.P.Shopen, T. ed. 1985. Language typology and syntactic description, 3 vols,
Cambridge: C.U.P.Sicking, C.M.J. 1971. Hoofdstukken uit de Griekse 'syntaxis', Amsterdam:
Athenaeum-Polak & Van Gennep. — . 1972. 'Heeft de Ilias een 'moraal'?', Lampas 5,444-64.Stechow, A. von 1984. 'Comparing semantic theories of comparison', Journal
of Semantics 3,1-77.Steinberg, D.D.; Jakobovits, L.A., eds. 1971. Semantics. An interdisciplinary
reader in philosophy linguistics and psychology, Cambridge: C.U.PStolz, B.A.; Shannon, R.S., eds. 1976. Oral literature and the formula, Ann
Arbor: Center for the Coordination of Ancient and Modern Studies, TheUniversity of Michican.
Strawson, P.F. 1950. O n referring', Mind 59,320-344 (Reprinted in Strawson(1971)).
— . 1971. Logico-Linguistic Papers, London 1971: Methuen.
Taglicht, J. 1984. Message and emphasis: on focus and scope in English, London:
Longman.Thompson, S.A. 1983. 'Grammar and discourse: The English detached par
ticipial clause', in Klein-Andrieu, ed. (1983), 43-66.
— . 1987. 'Subordination and narrative event structure', in Tomlin, ed. (1987),435-54.
Thompson, S.A.; Longacre, R.E. 1985. 'Adverbial clauses', in Shopen, ed.(1985), ii, 171-234.
Tomlin, R.S. 1983. 'Interaction of subject, theme and agent', Journal of Prag
matics 7, 411-432.
Tomlin, R.S., ed. 1987. Coherence and grounding in discourse, Amsterdam/Philadelphia: Benjamins (Typological Studies in Language).Traugott, E.C. 1985. 'Conditional markers', in Haiman, ed. (1985), 289-307.Traugott, E.C.; Ter Meulen, Α .; Snitzer Reilly, J.; Ferguson, CA., eds. 1986.
On Conditionals, Cambridge: C.U.P.
Visser, E. 1987. Homerische Versifikationstechnik Versuch einer Rekonstruk-
tion, Frankfurt: Peter Lang.
Vivante, P. 1982. The epithets in Homer. A study in poetic values, New Haven &London: Yale University Press.
Wallace, S. 1982. 'Figure and ground: the interrelationships of linguistic
categories', in Hopper, ed. (1982), 201-223.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 301/318
REFERENCES 291
Wendel, . 1935. Scholia in Apollonium Rhodium, Berlin: Weidman.Willcock, M.M. 1978. The Iliad of Homer. Books I- , London: Macmillan. — . 1984. The Iliad of Homer. Books XIII-XXW, London: Macmillan.Witte, . 1912. 'Zur Flexion homerischen Formeln', Glotta 3, 110-17.
Reprinted in Latacz, ed. (1979:109-17). — . 1913 'Horneros, ) Sprache', in. . Pauly - G. Wissowa, Real-Encyclopädie
der klassischen Altertumswissenschaft, Stuttgart 1893-1978.Zwarts, F. 1986. Categoriale grammatica en algebraïsche semantiek Een onder
zoek naar negatie en polariteit in het Nederlands, Diss. Groningen.
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 302/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 303/318
INDEX OF SUBJECTS
A fortiori argumentation, 79,80, 85,214,229
Acceptability scale, 95
Additive, 96,97,132
Substitutive, 96
Adhortation, 243
AdjectiveAnd time-stability, 126
As intensional term, 38
In analytic participial phrase, 5
Non-superlative, 126
Adverbial constituents
Dependency of, 112
Adverbial relation, 108
Adverbial semantic relation
And the study of 'participle + per', 109
Causal, 109
Conditional, 109
Expressed by participle, 10,23,108
Kinds of, 109
Adverbial subclause
Kinds of, 109
Adverbs
Scope of, 144
Aeolic, 24,26
Afterthought, 145
Alia, 234
Also, 51
And the same as-contexts, 51
Alternatives for focus constituent, 44-45,48,76,93,97,105,216,243
Scalar ordering of, 45
See also Focus constituent, Scale,
Scalarity
An, 72,217
With participles, 143
See also Ken
Anaphora, 113,115,118,124
Antonymic pairs, 104
Any
Existential vs. universal, 64
Negative polarity item, 54, 59,65
Positive polarity item, 65
Quantifier, 31,35,39
Aorist, 131
Vs. present, 130Apocope, 24
Apposition, 118,121
And identification, 118
And non-restrictive relative clause, 118
And syntactic independence, 121
Discourse-pragmatic motivation for,
118
Appropriateness, 37, 60,77
See also Reference, Speech act
Archaism, 20
Argonautica, 199
Arguments, 34,57,90,113,137,144
In participial phrase, 174
Aristarchus, 148
Aristotle, 103
Ascriptive sentence, 36,129,130
Aspect, 129,202
And the background-foreground dis
tinction, 147
And imperatives, 148
And time-stability, 130
Progressive, 147
Assertion, 32,40Positive vs. negative, 86
At least, 94-95
Attributive
Vs. predicative, 143
Vs. referential, 143
Auch nur (German), 47
And negative polarity, 56
Backgrounding, 72,131
Binary contrast, 28,69,104
Brevis in longo, 187,188,259
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 304/318
294SUBJECT INDEX
Bucolic diaeresis, 159, 160, 166, 167, 177,
191,247
Caesura, 80,103,165,167,181,253
Absence of, 170
Fluctuation between P- and T-caesura,
176,183,202, 224
Penthemimeral, 103,166,179,188,200
Trochaic, 103, 156, 166, 174, 175, 180,
182,187,200,224,247,267
See also Dactylic hexameter, MetricsCataphoric element, 118
Categorization
In cognitive psychology, 14
Logic-based theories of, 15
Platonic view of, 25
Wittgensteinian view of, 25
Category, 15
Discrete vs. continuous, 15
Internal structure of, 17
Linguistic, 17
Overlap between, 17
Relation between, 17
See also Continuum, Neutralization,
Prototype Theory
Checklist theory of category membership, 15
Circumstances, 118
Intension of, 114
Subject to scalarity, 113, 114, 115, 116,
119,121,123
See also Kai hos
Clausula, 159, 167, 169, 173, 191, 217, 245,
247,251,253,255,261,267,268
See also Bucolic diaeresis, Dactylichexameter
Co-extensive terms, 37,60,102
Co-ordination, 234
Cognitive psychology, 14,17,63
Command, 4,90,94, 95
Negative, 147
See also Directive speech act
Communicative intention, 43
Comparative, 49,79,81,101
And scalarity, 83
Context-dependent, 85
Morphosyntactic, 79
Comparative clause, 50,85
Comparative particle, 84
Comparative predicate, 83
Competence-performance distinction, 25
Complement clause, 108
Complete proposition, 90
Compromise, 94, 97,100
Compromise wish, 93,132
See also Per in wishes/commands
Concession, 9,10,49,80,107,120
And scalarity, 49
And the description of per, 49As a 'sense' of per, 10, 80,81,120
As opposition between scope sentence
and focus constituent, 49,81,116
Central meaning of 'participle + per',
107
Diachronic nature of, 14
Scalar nature of, 80,107,116,120
Concessive adverbial relation, 2,5,9,10,81
Concessive conditional, 2, 6, 205
And epistemic modality, 214
And factuality,220And indeterminacy, 208,227
And prototypicality, 207
And superlative fact, 48
And superlativeness, 208,222
Even if-type, 208,209,219,220,223,224
However much-type, 219,221-224,227
Kinds of, 207,209,223
Negation as lower value on the scale, 216
Vs. concessive, 206,215,223
Vs. simple conditional, 207,227,228
Concessive subordinator
Scalar origin of, 120,146
Concessives
Factual nature of, 206
Conditional, 35,42
And negative polarity items, 54,86,88
And the focus of even, 48,205,207,215
And the scope of even, 47,48,205
Concessive, 48,105,205
Contingency vs. indeterminacy, 63
Discourse function of, 234
Embedded vs. non-embedded, 62
Restrictive, 62Subjunctive, 34
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 305/318
SUBJECT INDEX 295
Connotation
Vs. denotation, 60
Context of utterance, 76, 85
Continuative propositional particle, 104
Continuum
Between various senses, 229
From concession to scepsis, 230-232
From extensional to intensional, 38
From formulaic to non-formulaic, 20
From prototypical to peripheral, 15-16
From scalar to concessive, 120From time-stable to time-unstable, 125,
129
From verbatim to rhythmical repetition,
159, 161
Conventional implicature, 41
Cooperative principle, 61
Copula, 129
Correlation
And concession, 119
Counterfactual, 58, 143
Dactyl, 165
Dactylic hexameter, 20, 165
Metric profile of, 165, 166,172,178
Quadripartite conception of, 200
Rhythmical structure of, 165,167
Rules and tendencies constituting the,
141,166
Tripartite instances of, 170,181,193
De, 213
And left-dislocation, 234
And topic marking, 234
Declarative sentence, 79,85,86,90,95,96See also Sentence-type
Definite article, 103
Degrees of comparison, 125,129
Demonstrative pronoun, 4,103,113
And correlation, 117
As extensional term, 38
Descriptive content, 76
See also Intension
Desirability scale, 91,94,244,247
Additive, 97
See also Per in wishes/commandsDiachronic descriptive linguistics, 13
Diachrony
As characteristic of epic diction, 20, 26,
74,146
Concession and, 14
Relation to synchrony, 120
Diathesis, 130
Direct speech, 82,103
Directive illocutionary force, 79,214
Directive speech act, 250
Discourse, 72,82
Dissuasion, 79,192,213
See also Command, Directive speechact
Double negation, 146
Dutch, 46,47
Eiper
And a fortiori argumentation, 214
And generic modality, 235
Concessive vs. non-concessive, 228
Diachronic status of, 226-227,232
Discourse function of, 210-211
In similes, 226
Non-superlative nature of, 209
Position with regard to main clause, 210,
218
Sceptic tone of, 229-231
Separate 'senses' of, 229
Elliptic clause, 50,63
Empês, 146,173,179,202
In participial phrases, 186
Metrical motivation for the use of, 172
Emphasis, 6,7,11
In the sense of 'highlighting*, 7-9
In the sense of 'intensification', 7,23Epic correption, 173
Epic diction
And linguistic research, 19,240
Diachronie nature of, 20,26,74,146,226
Equative sentence, 60
See also 'The same as-contexts'
Etymology, 24
Even, 4,27,30,40,41,76,77,95,96
And conditionals, 205
And conventional implicature, 42-43,
208And directive speech acts, 79
And focus particles, 40
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 306/318
296 SUBJECT INDEXAnd illocution, 43
And scope ambiguity, 46-47,56,64
And scope-dependent scale, 46
And strengthening, 101
And superlative fact, 78,123
As a positive polarity item, 53,121
As an inclusive focus particle, 44-45
In comparative expressions, 62,83
In indirect speech acts, 79
In negative contexts, 54
In negative polar contexts, 56Scalar implicature of, 45,48
Unspecified for scope-dependence, 46
Even if, 48
And directive speech acts, 213-214
Decompositional analysis of, 48, 214,
219
See also Concessive conditional
Event, 125,130
Exclusiveness, 91,93,97, 98
Property precluding scalarity, 93
Extension, 27, 37, 38, 45, 69, 76, 77, 84,96,225
As an inherent property of noun
phrases, 60
Vs. reference, 38
See also Reference, Scalarity exten-
sional
Extensional context, 60
Factive predicate, 42,58,143
FactuaHty, 33-34, 58-59, 70, 85, 94, 97, 111,
206-207,215,219-220,223
Family resemblance, 25Figure
Vs. ground, 50,63,101
Final subclause, 88
Focalization, 103
Focus, 44
Focus constituent, 40, 44, 45, 48, 51, 69, 75,
76,77,83,93,94,113,116,208,224,243
Intension of, 107
Vs. sister constituent, 62
Focus context, 49,269
And negation, 50,52
And the study of per, 51,52
Focus particles, 11,14,40
And at least- and at mart-meanings, 29
As positive polarity items, 53
Exclusive, 64
Inclusive vs. exclusive, 44,93
Vs. scalar particles, 40,242
Foreground
Vs. background, 101
See also Figure
Formula, 19,80
And 'redundant semantics', 163
And orality, 152,163And ordinary language, 157,164
And recurrence, 158
And substitution, 160,164
Definition of, 152-153,156,160,162
Flexible vs. fixed, 198
Flexion of, 203,255
Form separated from meaning, 154-156
Functional vs. structural approach to,
151-154,157,159,162-163,189,198,199
Juxtaposition of, 187,203
Linguistic account of, 151Localization of, 139,171,199
Metrical form of, 157-158
Modification of, 187,203
One-word formula, 162
Preserving influence of, 74
Prototypical vs. peripheral, 152
Semantic integration of, 19-20, 137,152,
189,192,204,217,241,244,248,250
Structural formula, 161-164,171
T1-formula, 156
T2-formula, 156,175,224
Use of in epic diction, 19,151
Functional Grammar, 105,145
Future, 33,58-59,90,143,233
Gar, 210,212
Ge, 7-9,18
Exclusive focus particle, 97,269
Generic present, 33,34,36,59
See also Modality generic
Generic subject, 137
German, 46,47
Gradience, 15,17
See also Continuum, Prototypicality
Ground
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 307/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 308/318
298 SUBJECT INDEXAnd enclitics, 169
Change of, 171-172
Delimitation of, 168
Localization of, 162,166-169
Negative factor in localization, 171
Vs. actual word, 169-171,203
Metrics, 20,103
And the form of formulas, 158
See also Dactylic hexameter
Mimesis, 103
Mimetic poetry, 103Minimal quantity, 55
Modal autonomy
And f actuality, 135
As a prototypical property of 'participle
+ per', 121-124, 133,135-136,165,193
Modal embeddedness
And 'participle + per', 133-135
Modality, 32,85
And referentiality, 34-35,59,104,222
Binary conception of, 33,104, 111
Deontic, 33,58, 59Epistemic, 58, 111, 144, 214
Generic, 4, 33, 70,72,138-139,212, 220,
235
Irrealis, 32-34,62,70,86,90-91, 111, 133,
217
Modal autonomy vs. embeddedness, 110
Possibility in the past, 70,72
Realis, 33-34,70,91, 111, 206
Scope of, 112,121,133,136,137
Modality verbs
Implicative, 143
Implicative vs. non-implicative, 59
Scope of, 146
Montague Grammar, 61
Mood, 33
Non-indicative, 33
Morphosyntactic categories, 130
Mycenaean, 26
Narrative passage, 81,203
Narratology, 82
Necessary and sufficient conditions, 25,31
Necessary condition, 124
Necessary enjambment, 267
Necessity, 33
Neg-raising, 102,146
Negation, 91
And 'participle + pef, 135
And irrealis modality, 35
And markedness, 64
And non-measurability, 127
And presupposition, 61
And scalarity, 52
And scale reversal, 54
And the reversal of truth-value, 54
As a lower value on a scale, 48,225As speech act, 104
Behavior of comparatives with regard
to, 50
External vs.internal, 53
In scope sentence, 53
Ontology of, 63
Scope of, 86,135
Strengthening of, 55
Negative
Vs. positive properties, 87
Negative assertionAnd negative polarity, 86
Mixed status of, 86
Negative polarity, 54,216,220
See also Negation, Negative polarity
items
Negative polarity items, 52, 54, 59, 86, 216,
237
As negation strengthened, 52,89
As scalar superlatives, 55,89
Distribution of, 52, 54
Excluded from wishes/commands, 91
Neutralization
Of differences between categories, 15,
17, 228,233,236,261
See also Category, Continuum, Overlap,
Prototypicality
Nog (Dutch), 46,104
Non-factuality, 33, 34,35, 59, 70, 85, 90, 94,
97,133,138,222
See also Modality irrealis
Noun-epithet formula, 155
And 'essential idea*, 154-156
And 'redundant semantics', 156, 158,171
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 309/318
SUBJECT INDEX 299And poetic quality, 201
Functional motivation for, 157
Metrical form of, 158
Traditional nature of, 152
Noun phrase
Definite, 41
Descriptive content of, 37,38
Extensional, 37, 76-77, 83, 96, 113, 253,
256
Indefinite, 34,59,104
Intensional, 37,69,76,96Non-referential, 35
Referential, 30,35,137
Referential vs. attributive use of, 60,76,
143
Nuclear predication, 114,116,123,146
Nucleus
Vs. periphery, 144
Ontology, 33,36,72,130
Ook maar (Dutch), 47
And negative polarity, 56
Optative, 58,91Oral performance and composition, 19, 20,
152,153,157,188,240
Oral-formulaic poetry, 19
Orality
Vs. formularity, 152
Vs. traditionality, 152
Orality test, 197
Oude, 77,252
And neg-raising, 102
Outer vs. inner metric, 165
See also Dactylic hexameter, MetricsOverlap between categories, 17, 147, 233,
237
See also Category, Continuum,
Neutralization, Prototypicality
Paradigmatic relation, 29,39,40,44,48
Participle
Adjectival (vs. verbal), 125,130,139
Analytic, 5,121,125,129,173
And aspect, 143
And complementation, 108
And syntactic independence, 110,112And time-stability, 121
As a non-sentential element, 109
Attributive, 138
Circumstantial, 5, 9-10, 108, 114, 137,
138
Classification of, 109
Concessive relation 'expressed' by, 110
Discourse function of, 109,143
Modally autonomous vs. modally em
bedded, 110-112
Perfect, 177
Predicative vs. attributive, 108Synthetic, 5,121,125,130,173
Unspecified nature in comparison with
adverbial subclause, 10,109
Vs. adverbial subclause, 108
'Participle + per
Alleged causal sense of, 134-135
And 'redundant semantics', 172
And commands, 135-136,203
And negation, 135
And semantically empty substantives,
174,177,186And the semantic integration of for
mulas, 137
As a formula, 139,157,162,164,165,170,
191
Competing with scalar superlatives, 139-
140
Factual nature of, 121,135-136,206
Function of in epic diction, 151
Linguistic vs. formulaic approach to, 107
Localization of, 141,166,170-173
Metrical form of, 151,164
Modally embedded instances of, 121
Prototypical properties of, 120-124,140
Scalar vs. concessive use of, 120
Vs. concessive conditional, 223
Per
And 'redundant semantics', 176
And German auch, 11-13,71
And negation, 127
And non-scalar focus particle, 257
And scalar superlatives, 69,250
And strengthening, 258
As a prefix in Latin, 13As negation strengthener, 89
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 310/318
300 SUBJECT INDEX
Collocation with empês, 179
Collocation with kai, 12,23
Collocation with mala, 177-178,185-186,
222
Collocation with oude, 256
Concession as 'sense' of, 10, 80-81,117,
120
Concessive vs. non-concessive, 2
Delimitation with regard to ge, 7-9, 97,
261
Delimitation with regard to kai, 10, 12,27, 67, 96,117,205,225
Determinative use of, 7-8,243
Diachronic description of, 3,21,98, 225
Diachronic development of, 14,107
Diachronie development parallel with
te, 74
Diachronie relations between uses of,
98,225,262
Diachronie status of use as scalar super
lative, 74
Dislocation of, 252-256Distribution of, 1
Enclitic, 1,4,51,170,177,253
Etymology of, 12,14,71
Homeric vs. post-Homeric, 1-2, 51, 63,
243,259, 263
In comparative expressions, 84
In conditionals, 6,9,205 ff.
In hos tø-clauses, 73-74,250
In negative polar contexts, 88
In non-restrictive relative clauses, 4, 9,
75,107,113
In participial phrases, 4,9,107 ff.
In 'the same ay-contexts', 6, 51, 75, 259,
268
In wishes/commands, 4, 91-98,131,244,
247, 252
Intensive use of, 7-8,13-14,24,71,85,90,
132,134,247,269
Limitation as 'sense' of, 105
Original meaning of, 13-14, 71, 85, 90,
132,134,247,269
Peripheral instances of, 6,240 ff.
Prototypical properties of, 18Range of the description of, 7,10
Scalar vs. non-scalar, 3,51
'Subordinator + per', 1-2, 23, 51, 263-
265
Substituted for byker, 174
Synchronic description of, 3,98,107
Perì, 13
Intensive use of, 13,101
Peripheral, 19
Peripheral instances, 62,99,140,239 ff.
And classification, 240
And diachrony, 261And divergence, 240,249,256
And epic diction, 240-241
And integration of formulas, 242-244,
246
And metrics, 241,257-258,261,268
And post-Homeric per, 262-265
And redundant semantics, 241-242
And strengthening, 257
See also Prototype Theory, Residue
Persuasion, 79,104
See also Directive speech act, Dissuasion
Polarity, 52
Negative, 47,87-88
Positive, 52,86-87,95
Polarity contexts, 35,52
Polarity item, 52
Politeness formula, 194
Polysemy, 24,93
Positive polarity items, 52
Possibility, 33
Posthomerica, 199
Pragmatics
Vs. semantics, 62
Predicate, 90
Adjectival vs. substantival, 121,125
Nominal, 129
Substantival, 125,128,129
Verbal, 125
Predication
Nuclear, 57
Presupposition, 40
Existential, 61
Semantic, 61,237Suspension of, 237
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 311/318
SUBJECT INDEX 301Pronominal adverb, 114
Proper name, 70,76-77
As extensional term, 38
Focus constituent of kai, 77,253
Scalar use of, 76
Proposition, 33,57
Propositional content, 42
Prototype Theory, 14-17,29,38,98,120,140,
146,189,229,230,239,257
Prototypical, 25
Vs. peripheral instances, 14,17,18, 25,124,149,239
Prototypical properties, 16, 18, 31, 70, 120,
124,135,146,190
Of'participle + per'. 120
Prototypicality, 14-15,19,20,59,256
And diachrony, 124,146
And predictability, 140,257, 258
And the semantic integration of for
mulas, 189-195
Degree of, 25
QuantifierExistential, 31,35,64
Universal, 31,35,64
Redundance, 20
Redundant semantics, 241,267
And metrics, 241
And strengthening, 241
Reference
And felicity, 77
Connection with modality, 34
Context-dependent, 76
Spatial vs. temporal, 34
Relative clause
And concessive interpretation, 80,107
And intension, 77
Concessive, 2,4
Non-restrictive, 4,28,68,72,75,252,253,
257
Restrictive, 59,75,88
Scalar, 75
Relative subordinator, 103
Repetition
Metrical, 159-162On the level of category, 160-164,171
Verbatim, 159-161,186
Residue, 15,18, 240,257, 261
Nature of the - of per, 19,240
See also Peripheral instances, Prototype
Theory
Rhetorical question, 54
Romance, 59
Rules
Linguistic, 25
Satisfactoriness, 244
See also Acceptability
Scalar particles, 27,39
And high points of scales, 30
Subclass of focus particles, 30,40
Scalar superlatives, 31,45,208,250
Affinity with any, 31
And illocution, 43
And intensionality, 39
And metrical localization, 184
And morphosyntax, 35-36,69
And negation, 53
And negative polarity, 55 Aadper, 35,69
And scope independence, 45
And stress, 49
And superlative fact, 31,69
As focus constituent of even, 62
Competing with 'participle + per 3 , 139-
140
Distribution of, 32,35,81
In hos te-clauses, 73
Neutrality of, 82
Non-referentiality of, 31, 35Scalarity, 3,14,28,39
And adverbial relation, 110
And comparatives, 83,84
And concession (diachronic), 81, 116,
117,120,124,131,146,254,262
And concession (synchronic), 81, 120,
124,253
And negative polarity items, 55
As a prototypical property of per, 27,
242,246,257
Extensional, 39-40,205,208,225
In wishes/commands, 131
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 312/318
302SUBJECT INDEX
Intensional, 71-72, 81-82, 116,124, 205,
208, 221,225, 253,256
Positive polar, 86
See also Scalar particles, Scale, Superla-
tiveness
Scale, 28
And at least- and at most-meanings, 29
And implicative, 29
As a continuum between two extremes,
29
High point of, 29, 45, 69,208, 246Implicated by scalar particle, 45
Semantic feature as identity of, 29
Scale reversal, 53-54,56,87
And markedness, 55
And negative polarity, 55
Scholia, 148
Schon (German), 46
Scope
Of focus particles, 44
Wide vs. narrow, 46-47,64,86
Scope ambiguityAnd conditionals, 47
And negation, 47,62
Scope-dependence, 45,69-70,76,84
And extensionality, 45,208
Scope-independence
And intensionality, 45,208
Scope sentence, 44-45,47,80,83,87,114,205
Semantic adverbial relation
Expressed by participle, 108
Semantic integration of formulas
And peripheral instances, 241
SemanticsVs. pragmatics, 62
Sense, 10
Vs. reference, 60
Sentence-type, 79
Declarative, 85-86,144
Imperative, 85-86,144
Interrogative, 85-86
Showing
Vs. telling, 82
Simile, 226
Diachronie status of, 226Extension of, 212
Situational knowledge, 76
Sogar (German), 46,47,54
And positive polarity, 64
Speech act, 43,79, 233
Direct vs. indirect, 79
Directive, 79 ,81
Felicity of, 58,192
See also Illocution
Spondee, 165
See also Dactylic hexameter
State of affairs, 30,33,40,57,70,86,109,113,116,118,123,212
Stoics, 60
Strengthening, 9,12,20,26,62,116,117,140,
172
And redundant semantics, 140,241,257
As a function of kai, 10
As a function of per, 9,10,109,257
Stress, 40,44
Subclause
Adverbial, 108
Complementary, 108Subordinating conjunction, 1,109
Substitution, 29,44,80
And the recurrence of formulas, 160,
164
See also Alternatives for focus con
stituent
Sufficient condition, 31
Superlative
And negative polarity, 54
Attributive, 37
In scalar statements, 30
Morphosyntactic, 78Negative, 55
Positive, 55
Pragmatic, 32,78,246
Referential, 37
See also Intension, Scalar superlatives,
Scalarity
Superlative fact, 31-32,43,48 ,69,71,78,82,
113,123,206,213,246
As a technical term, 44
See also Illocution
Superlative property, 76-78,84,113
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 313/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 314/318
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 315/318
INDEX OF NAMES
Adkins, A. W.H 204Altmann, H. 57Austin, N. 153Auwera, J. van der 57, 63, 64,105
Baker, G.P. 25, 61Bakker, EJ. 2, 6, 26, 34, 51, 60, 61,
62, 63, 105, 145, 147, 203, 235,236,237,269
Bakker, W.F 148Bal, M. 103Beekes, R.S.P. 160, 166, 167, 168,
169,200Bennett, J. 61, 62, 63,104
Benveniste, E. 57,101Bolkestein, A.M. 62,101,102Booth, W.C. 103Brugmann, . - Thumb, A. 24Buck, CD 24
Carlson, G. 64Chantraine, P. 7, 9, 23, 24, 25, 26,
144,145,148,203,235Combellack, F.M. 188
Cruse, D.A. 24Davison, A. 58,59,64,105Denniston, J.D. 7, 8, 10, 13, 14, 23,
24, 85, 90, 105, 132, 135, 204,236,237,243,268,269
Dik, S.C. 57,105,145,198,234,235Donelian, K. 60,76,143Ducrot, . 57,102
Ebeling, H. 84,134
Edwards, M.W. 201
Fauconnier, G. 32,53,54,57,58,61,62,64
Fillmore, C.J. 25
Fleischman, S. 58,59,143Ford, CE. 234Fox, B. 143Fraenkel, H. 7, 11-14, 23, 71, 200,
204Fraser, B. 61Frege, G. 37,60, 61Frisk, H. 24
Gazdar, G. 57
Gehring, A. 277Genette, G. 103Givón, T. 15,16,25,32,41,58,59,60,
61,62,63,64,101,104,105,143,145.147.234
Goodwin, W.W. 23, 105, 108, 110,143.147.235
Grice, H.P. 41, 61,237
Haack, S. 60
Haas, W. 197Hacker, P.M.S. 25,61Haiman, J. 24, 63,233,234,237Hainsworth, J.B. 162,198,199Haudry,J.145Hawkins, J.A. 59Hoeksema, J. 63, 64Hoekstra, A. 74, 103, 162, 169, 171,
197,200,201,202,203
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 316/318
306 AUTHOR INDEX
Hoffmann, M.E. 105
Hopper, P.J. 25,58,59,63,101
Horn, L.R. 40, 52,57,58,59, 61, 64,
89,102,105,237
Ickler, N.L. 234
Ingalls, W.B. 199,200
Jacobs, J. 57, 61,64
Janko, R. 74,106Jong, I.J.F. de 103
Kartonen, L. 41, 42, 43, 44, 45, 46,
57, 61, 62
Kempson, R.M. 61, 64
Kiparsky, 58,61,143
Kiparsky, P. 58,61,143,198,199
Kirk, G.S. 170,197,200,267
Kirsner, R.S. 24
Klein-Andrieu, F. 24Kneale, W. & Kneale, M. 60
König, E. 25,40,41,42,45,49,57,61,
62, 63, 64, 102, 105, 146, 207,
216,228,233,235,237
Kühner, R. - Gerth, . 7, 8, 23, 24,
72,103,131,143,145,146,147,
235,236,237
Ladusaw, W. 64
Leaf, W. 134,145,269Leech, G.N. 102
Lehmann, Chr. 105,145
Levinson, S.C. 57, 61, 62,64,237
Lidell, H.G. & Scott, R. 237
Longacre, R.E. 63,235
Lord, A.B. 197,199,200
Lyons, J. 24, 36, 57, 58, 60, 62, 73,
102,104,129,144
Matthews, P.H. 57,144,145Meister, . 154,158,201
Mffl,J.S.60
Minton, W.W. 162,199
Monro, D.B. 7,8,13,24,235,243
Monteil, P. 7,8,9,23
Murray, G. 267
Nagler, M.N. 235
Notopoulos, JA. 199
Oguse, A 9, 23, 110-112, 121, 144,146.148.224.236.268
O'Neill Jr., E.G. 159,160, 165, 166,
168-171, 199, 200, 201, 202,
203,267
Paardekoper, P.C. 64
Packard, D.W. 199
Palmer, RR. 58,144
Parry, M. 25, 26, 152-163, 172, 173,
187, 197, 198, 199, 200, 201,203,267
Peabody, . 197
Peters, S. 41,42,43,44,45,46,57,61,
62
Pinkster, H. 145
Porter, H.N. 200,201
Prendergast, G.L. 201
Quine, W.V.0.61
Quirk, R. et al. 63,233Raalte, M. van 165, 167, 168, 200,
201
Ramsey, E. 101,234,235
Rescher, N. 144
Rieu, E.V. 3,235,267
Rijksbaron, A 143,147,234,236
Rosch, E.H. 15
Ruijgh, J. 23,24,26,57,72,74,101,
102, 103, 106, 144, 147, 148,234.235.236.268.269
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 317/318
AUTHOR INDEX307
Russell, . 64Russo, J.A 161-164, 197, 199, 2
203
Schmerling, S. 64Schoorl, S. 60Schwyzer, E. - Debranner, A 24Searle, J. 102Seuren, P.AM. 52, 61, 64,102Shipp, G.P. 226,236Sicking, C.M.J. 144,148,204Stechow, A. von 63Strawson, P.F. 60, 64
Thompson, S.A 25,59,63,101,143,234,235
Tomlin, R.S. 57Traugott, E.C. 57,233
Vandeweghe, W. 57Visser, E. 197,198,201Vivante, P. 201
Wallace, S. 58, 63,101Wendel, . 148Willcock, M.M. 148Witte, . 154,158,201,202,203
Taglicht, J. 57Zwarts, F. 64
8/16/2019 Linguistics and Formulas in Homer
http://slidepdf.com/reader/full/linguistics-and-formulas-in-homer 318/318