AMSTERDAM STUDIES IN CLASSICAL PHILOLOGY

EDI TORS

A L B E R T RIJKSBA RON
I R E N E J.F. DE JON G HARM P IN K S TE R

V OL UME SEVEN

P R E VI O U SL Y PUBL ISH ED

1. A. Rijksbaron, G r a m m a tic a l O b s e r v a tio n s o n E u r i p i d e s ' B a c c h a e . 1991.

2. R. Risselada, I m p e r a tiv e s a n d o t h e r D ir e c tiv e E x p r e s s io n s in L a tin . A S t u d y in th e
P r a g m a tic s o f a D e a d L a n g u a g e . 1993.
3. G. Wakker, C o n d itio n s a n d C o n d itio n a ls . A n i n v e s t i g a t i o n o f A n c i e n t G r e e k .
1994.
4. C. Kroon, D i s c o u r s e P a r t i c l e s in L a tin . A S tu d y o f nam, enim, autem, vero a n d
at. 1995.
5. H. Dik, W o r d O r d e r in A n c i e n t G r e e k . A P r a g m a tic A c c o u n t o f W o r d O r d e r
V a r ia tio n i n H e r o d o tu s . 1995.
6 . J.E. v.d. Veen. T h e S i g n i f i c a n t a n d th e I n s ig n ific a n t. F i v e S tu d ie s in H e r o d o t u s ’
V ie w o f H i s t o r y . 1996.
 NEW APPROACHES TO
GREEK PARTICLES

P R O C E E D I N G S OF THE C O L L O Q U I U M H EL D IN
A M S TE RD A M , JA N U A R Y 4-6, 1996,
T O HONOUR C.J. RUIJGH
ON THE OC CA SI O N O F HIS R E T I R E M E N T

EDITED BY
ALBERT RIJKSBARON

J .C - GIEBEN. PU B LISHER
AMSTERDAM 1997
No part of this book may be translated or reproduced in any form, by prim, photoprint,
microfilm, or any other means, without written permission from the publisher.
Printed in The Netherlands / ISBN 90 5063 097 9
 PREFACE

On December I, 1995 Kees Ruijgh retired from the University of Amsterdam, after a
long and fruitful career devoted to the study of Greek, first as a junior researcher and
teacher, and from 1969 as professor of Greek linguistics. During his career, a great
variety of subjects attracted his scholarly attention, for which the reader may be
referred to the ‘Avant-propos’ of Ruijgh’s S c r i p t a M i n o r a /, and to the ‘Biblio­
graphie complète’ printed in S c r i p t a M i n o r a I I . Among these subjects, the Greek
particles take a place of honour. We need only mention here his monumental A u to u r
d e τ ε é p i q u e of 1971, which, besides many other things, is an in-depth study of τε
and related particles. So when the Department of Classical Philology started organiz­
ing a colloquium in his honour and was looking for a general theme for this meeting,
it seemed to us that the particles were a particularly suitable subject, the more so
because they have been rather neglected since the publication of Denniston’s T h e
G r e e k P a r t i c l e s , a verdict which does not apply to, say, verbal aspect, to name only
one other suitable subject.

The colloquium took place in Amsterdam on January, 4-6 1996, and brought together
some fourteen speakers, representing five different European countries and several
academic generations. Two lectures presented at the colloquium are absent from the
present book, since—much to their regret—Frangoise Létoublon (Grenoble) and
Michael Meier-Briigger (Hamburg/Berlin), due to other obligations, were not able to
send in a worked-out version of their lecture.

it is our pleasure to thank a number of institutions and persons for their support.
Financially, the colloquium was made possible by grants from the Faculty of Arts
and the Institute of Mediterranean Studies of the University of Amsterdam, the
Amsterdam University Association, the Royal Academy of Arts and Sciences and the
Publisher, J.C. Gieben. Organizing the colloquium would have been far more diffi­
cult without the untiring efforts of Ruijgh’s former student, Hotze Mulder. The
Introduction has much benefited from the criticism by Irene de Jong and Rodie
Risselada.

It will be generally agreed that Kees Ruijgh, to use one of his own favourite phrases,
‘a bien mérité de la linguistique grecque’. We hope that the collection of articles
assembled in the present book will be up to his merits.

Amsterdam, November 1997 Albert Rijksbaron

 CO N TEN TS

Preface........................................................................................................................... v
ALBERT RUKSBARON
Introduction................................................................................................................... 1
YVESDUHOUX
Grec écrit et grec parlé. Une étude contrastive des particules aux Ve-IVe siècles......15
Anna Morpurgo Davies
Particles in Greek Epigraphical Texts. The Case of Arcadian....................................49
Louis Basset
Άλλ' έξόλοισθ’ αύτω κοαξ. Réexamen des emplois de άλλα à la lumière de
l’énonciation dans les G r e n o u i l l e s d’Aristophane.....................................................75
S.R . SLINGS
Adversative Relators Between PUSH and POP........................................................ 101
BERNARD JACQUINOD
Sur le rôle pragmatique de καίτοι............................................................................ 131
A. MARIA VAN ERP TAALMAN KlP
η γάρ in Questions.................................................................................................... 151
C .M .J. SICKING
Particles in Questions in Plato..................................................................................157
IRENE J.F . DE JONG
Γάρ Introducing Embedded Narratives....................................................................175
ALBERT RUKSBARON
Adverb or Connector? The Case of κ α ι... δέ..........................................................187
G erry W akker
Emphasis and Affirmation. Some Aspects of μην in Tragedy.................................209
INEKE SLUITER
Parapleromatic Lucubrations....................................................................................233
P aul Wathelet
Les particules κε(ν) et αν dans les formules de l’épopée homérique..................... 247
F .M .J WAANDERS
Particulars of Some Proto-Indo-European Particles................................................269
Indices...................................................................................................................... 275
 INTRO DUC TIO N

A l b e r t R ijk s b a r o n
Universiteit van Amsterdam

‘articles, particles, prepositions, auxiliaries

... act as policemen and direct each of the
other words to its proper place in the brain
o f the hearer so as to facilitate orderly
understanding’—Otto Jcspcrscn

In the early 19th century, the city of London was the first city to have public gas­
lighting. As a result, one school of historians claims, it took a considerable time
before it got electric light. This process is regarded as a prime example of the work­
ings of the so-called ‘Law of the Retarding Lead’.
In 1934 Denniston’s epoch-making T h e G r e e k P a r t i c l e s appeared.1 It was
especially after the publication of the second edition, in 1954, which contained an
invaluable i n d e x l o c o r u m (due to his wife), that ‘Denniston’ became one of the
indispensable tools of the Greek scholar. I submit that most, nay all, Greek scholars,
if they were allowed to take just ten professional books to the proverbial deserted
island, would include Denniston, together with Kiihner-Gerth and Liddell-Scott-
Jones. The book was so good, in fact, and so much ahead of what was done for other
languages, that it acquired the position of London’s gas-lighting: in Greek linguistics
Denniston’s monograph simply became th e standard reference book, and for a long
time there must have been a widespread feeling that improving upon his treatment
was not feasible and a waste of time. At least, this can be gathered from the
omnipresence of his name in commentaries, often in the form of simple references,
and from the surprisingly small number of detailed particle studies published after
1934 up to the seventies.1 2
The excellence of T h e G r e e k P a r t i c l e s resides mainly in the choice and the dis­
cussion of the examples. In fact, on encountering a particle which arrests our atten-

1 Incidentally, it is worth noticing that in 1950 a book by C.D. Anderson appeared, entitled ‘The
Elementary Particles of Physics’. Some twenty years earlier the outlines of this new branch of
physics had become visible, but in 1950 apparently both ‘elementary’ and ‘physics’ still had to be
added to make clear what kind of particles the book was about. I shall resist the temptation of
ascribing this fact to the publication, also in the thirties, of Denniston’s Panicles, but there can be
no doubt that in the thirties and forties the word ‘particle’ primarily still had a linguistic sense.
Nowadays ‘particle’ simply stands for ‘elementary particle’, as appears e.g. from a publication like
Megascience: Particle Physics, published in 1995. (Not that the presence of ‘Megascience’ would
have induced us to believe that the particles of language were concerned.)
2 Ruijgh’s Autour de τε épique, published in 1971, was the first major monograph on Greek
particles since ‘Denniston’.
 ALBERT RIJKSBARON

tion, we all tend, I think, to look up in the index whether Denniston discusses that
particular instance, and feel satisfied when he does, the more so because he usually
provides us with a sensible and sensitive remark. And fortunately he discusses an
incredible number of instances; as he himself puts it: ‘The reader should be enabled
to b a th e in examples’ (Preface to the first edition, p. vi; italics Denniston). While the
‘fast food’ use of Denniston undoubtedly accelerates our reading-pace, things are
less simple when it comes to finding the precise meaning and the syntactic function
of a given particle, especially in particle combinations. On this count. T h e G r e e k
P a r t i c l e s is all too often disappointing and inadequate. A typical illustration of this
inadequacy is a statement like the following: O n the other hand, Bäumlein, Kühner,
and others deny that γάρ in άλλά γάρ is ever causal, and interpret it everywhere as
“adverbial”.’ (p. 100). Apparently for Denniston ‘causal’ and ‘adverbial’ exclude
each other. Strange though this opinion may seem, it is in line with Denniston’s use
of ‘adverbial’ in his Introduction; on p. xxxix he tells us that he uses ‘adverbial’
notably for ‘particles of emphasis and nuance, since they are in most cases naturally
translated by adverbs...’ And because, for him, γάρ is not a particle of ‘emphasis
and nuance’, it cannot be an adverb in the combination άλλά γάρ. In the last resort, it
remains unclear what is the status of γάρ in this combination. Similar problems con­
cerning the syntactic status of particles are involved in a number of other combina­
tions, e.g. και (...) γάρ and και (...) δέ. This, in tum, is presumably connected with
the fact that Denniston discusses such combinations in an in itself consistent, but
nevertheless rather peculiar way; thus, all combinations of καί are discussed under
the other particle involved (και (...) γάρ under γάρ, καί (...) δέ under δέ, etc.).
Probably as a result, Denniston never investigates s y s t e m a t i c a l l y the function of καί
(and thus that of the other particle) in these combinations. That Denniston sometimes
must have felt quite helpless, when confronted with syntactic problems, is apparent
from a note on page xliii: ‘The line between connectives and non-connectives cannot
be rigidly drawn. Thus ούν in Homer, although it has n o t y e t developed a connective
function, shows in μέν οΰν a t e n d e n c y to develop one. γε, and in a more marked
degree μέν γε, mitigate to s o m e e x t e n t the harshness of an asyndeton : while γοΰν in
the ‘part-proof’ usage is a l m o s t a full connective, μέν, again, o c c a s i o n a l l y a p p e a r s
to have a <?uayt'-connective force.’ (Italics A.R.). One hopes there is some room for
greater exactness, here.
On the syntactic side, then, much remains to be done. Needless to say, syntactic
refinements are inseparable from refinements on the semantic and, I should add, on
the pragmatic side. As to the latter branch of linguistics, over the past twenty years
much work has been done concerning other languages than Greek, and chances are
that this work may shed some new, ‘electric’, light on the Greek particles. At least,
that is the assumption underlying six of the articles collected in this volume (those by
Basset, Slings, Jacquinod, Sicking, Rijksbaron and Wakker).3 The ‘New Approach­
es’ mentioned in the title of this book are, then, for a large part pragmatically oriented

3 Needless to say, the degree to which these authors subscribe to the approaches mentioned below
may vary.
 INTRODUCTION

approaches. Within these approaches two main directions can be discerned, that are
connected with the names of the discourse analysts E. Roulet, who initiated the so-
called ‘Geneva model’, and O. Ducrot.4 By a happy coincidence, their work and that
of their followers has recently been summarized and integrated in the thesis of Caro­
line Kroon on Latin discourse particles; she has added, moreover, a number of
insights of her own. I shall therefore briefly describe those elements of Roulet’s,
Kroon’s, and Ducrot’s work that have been used to a greater or lesser degree by the
authors mentioned above, and that to my mind are of special interest for a satisfactory
study of the pragmatics of particles. These are: Kroon’s ‘descriptive model of dis­
course’ (mainly to be found in Chapter 4 of her thesis. In the present collection of
articles this model is used most extensively by Wakker), and Ducrot’s argumentation
theory, which is represented notably by Basset.

Following Roulet, Kroon assumes that every discourse is not just a series of linearly
ordered speech acts, but consists of hierarchically ordered units. To ensure a success­
ful communicative act, these units in some way or other must cohere. Particles are an
important means to signal coherence. A further assumption is that coherence is
established at three different discourse levels: the representational, the presentational,
and the interactional level. The first relates to the world as it is represented through
language. While particles as such, unlike e.g. nouns and verbs, do not refer to this
world (have no referential meaning), they may function as devices to connect the vari­
ous entities and events that make up the represented world, indicating that they
semantically belong together. Some Greek examples are αλλά and καί (in their con­
nective use). The presentational level concerns the ways in which a speaker organizes,
or ‘stages’, to use Kroon’s term, the information he wants to communicate to the
hearer. Particles operating at this level divide the discourse into central and peripheral
units, mark explanatory and digressive units, help the hearer keep track in the dis­
course, briefly, guide the hearer through the discourse.5 Greek examples are οΰν and
δέ. Finally, the interactional level concerns the ways in which the, cognitive and emo­
tional, relationship between a hearer and a speaker in a particular communicative sit­
uation is shaped. Particles at this level involve notably the attitudinal (also called
modal) particles, i.e. particles that specify the attitude of the speaker with regard to the
information he is proffering, e.g. his commitment to its truth, or his belief that this
information is of special interest for the hearer. Greek examples are που ‘perhaps’,
and τον ‘I assure you that, you take it from me that’.
Following, and slightly modifying, Ducrot, Basset distinguishes four constitutive
elements in any speech act: a) elements that are presupposed, b) the discourse theme,
which must have been agreed upon by both speaker and hearer, c) elements that are
posited (‘le p o s é ’), also called rhematic elements (new information), and d) elements

4 For bibliographical details I refer to the the articles by Basset, Slings and Wakker.
5 To avoid a possible misunderstanding, I should add that units that in some way are related at the
presentational level, will normally, of course, also be related at the representational level. Thus,
while Greek δέ marks a new discourse unit, this unit is related qua content with the previous
unit(s).
 ALBERT RIJKSBARON

that are not overtly present in the ‘posited’ elements but are understood or hinted at
(‘le s o u s - e n t e n d u ') . Ducrot and his collaborators have used this speech act model
with considerable success in analysing e.g. French m a i s , and Basset, and to a lesser
extent Jacquinod, follow their lead in their analyses of αλλά and καίτοι.

The ‘newness’ of the approaches brought together in this book is by no means con­
fined, however, to pragmatics. This is amply demonstrated by the articles of Duhoux
and Morpurgo Davies that open the book.
In ‘Grec écrit et grec parlé’, Yves DUHOUX has undertaken the rather formidable
task of investigating in great detail, and supported by a full statistical apparatus, the
distribution of the most frequent particles across the various genres and authors and,
more importantly, across the various types of discourse, with the emphasis on dia­
logic as against non-dialogic discourse. His point of departure is the wide-spread
belief, which for some is rather a near-certainty, that in everyday speech particles were
far more common than in the written language, and that this situation is reflected in
the dialogic parts of Greek literature. As Denniston puts it (p. lxxii): ‘It cannot be
doubted that Greek conversation was full of particles: at moments of excitement and
emotional tension the dialogue of tragedy and comedy fairly bristles with them.’
Note that Denniston here, in a rather amazing move, simply equates ‘conversation’
with ‘the dialogue of tragedy and comedy’, the—implicit—assumption being that the
dialogue of drama contains more particles than the rest of our texts. Some pages
further down (p. lxxv) he restates this opinion (‘I have observed that Greek drama
reproduces, as far as one can tell, the free use of particles in everyday speech’) and
adds: ‘Particles are on the whole, I think, more often employed in comedy than in
tragedy.’
One of the important findings of Duhoux, based upon a thorough comparison of
a corpus consisting of Plato’s M e n o , A p o l o g y o f S o c r a t e s and P r o t a g o r a s ,
Xenophon’s S y m p o s i u m , Aristophanes’ T h e s m o p h o r i a z u s a e and Sophocles’ O e d i ­
p u s a t C o lo n u s , is that, on the contrary, particles, taken as a whole, are le s s frequent in
the spoken parts of his corpus. In his own words (p. 39): ‘... il est sans doute inexact
de penser que les particules étaient spécialement fréquentes dans la langue parlée.
C ’est l’inverse qui est probablement vrai: les particules étaient davantage utilisées à
l’écrit.’ He sees a confirmation of this conclusion in the fact that the dialogic parts of
Xenophon’s S y m p o s i u m contain less particles than the non-dialogic parts. Duhoux
has also found that a number of particles are always more frequent in non-dramatic
texts, among them the coordinating conjunctions ή, καί and μέν. This, in turn, sug­
gests that in drama asyndeton is more frequent than in the other text-types; as a con­
sequence the language of everyday speech, if it is, indeed, reflected in that of drama,
must also have used asyndeton on a large scale.
As to the belief that particles are more frequent in comedy than in tragedy, this,
too, proves to be incorrect: there is no significant difference, neither in the dialogic
nor in the non-dialogic parts, between the T h e s m o p h o r i a z u s a e and the O e d i p u s a t
C o lo n u s .
 INTRODUCTION

The lesson to be learned from Duhoux’ investigation is that one should not treat
the Greek particles as a monolithic block: general statements like ‘particles occur
more frequently in the spoken parts of Greek texts’ should be avoided, and be
replaced by statements like ‘the particles x, y and z are more f r e q u e n t e t c . Also,
the idea that drama is especially suited to give us an insight into the language of
everyday speech lacks a solid foundation. I am even inclined to believe—but Duhoux
would perhaps not share my view—that the very idea that spoken language can be
extrapolated from Greek in its written form, has to be abandoned.

In her article ‘Particles in Greek Epigraphical Texts’, Anna MORPURGO DAVIES

takes issue with another common belief, of an equally generalizing nature, namely
that ‘Greek is unbelievably rich in particles.’ Denniston, for instance, writes that
‘[differences in dialect play a certain part, b u t... perhaps a rather smaller part than
we should have expected’ (p. lxx). Having observed that Denniston here only has lit­
erary dialects in mind, Morpurgo Davies breaks new ground by focusing on the use,
and indeed the very existence, of particles in inscriptions in general and in Arcadian
inscriptions in particular. She notes at the outset that ‘outside literature most particles
are not attested or are badly attested.’ Thus, in a corpus of 500 verse-inscriptions
written before 400 BC ‘only καί, δέ and τε or τε ... καί occur reasonably frequent­
ly’ (p. 51). Arcadian presents a similar picture; here, only κάς/καί and δέ occur with
some frequency. After a discussion of κάς/καί she arrives at the conclusion that the
form καί is probably a borrowing from other dialects. This is an important conclu­
sion, since it invalidates another common belief, viz. that particles are a closed class,
which can only change, if at all, under internal, not under external influence.
The bulk of her paper is devoted to the function(s) of δέ in the Arcadian inscrip­
tions. Making use of the pragmatic notions developed by E. Bakker, she observes
that δέ gradually turned from a ‘seriously’ adversative particle, as she puts it, into a
‘transitive’ or ‘continuative’ particle that serves to establish cohesion in a text, while
at the same time marking thematic and other shifts, a function well-known from other
texts. Now the earlier exclusively adversative value is something of an oddity, espe­
cially in view of the fact that in Mycenaean the ‘transitive’ use is already well-estab­
lished. She proposes an explanation by arguing that in early Arcadian δέ contrasted
systematically with asyndeton: because asyndeton marked the transition to another
topic in a neutral way, δέ could acquire a ‘clearly polarized adversative value’ (p. 67).
She further argues that the change to the ‘continuative’ function was, again, largely
due to external influence. In her conclusion she defends her emphasizing the role of
interdialectal borrowings by pointing to recent studies of living languages, that show
that such a procedure definitely has to be reckoned with.

The paper by Louis BASSET is the first of a trio that deals with adversative relations.
Taking the seminal work of Oswald Ducrot on French m a i s as his starting point, he
investigates all uses of άλλα (with the exception, though, of adverbial άλλα and
combinations like ού μην αλλά) in Aristophanes’ F r o g s , his ultimate goal being to
 ALBERT RIJKSBARON

show that άλλά is semantically uniform, and that the differences result from the vari­
ety of the elements preceding άλλά. As noted above. Basset distinguishes four con­
stitutive elements in any speech act; for convenience’ sake I repeat them here. There
are: a) elements that are presupposed, b) the discourse theme, which must have been
agreed upon by both speaker and hearer, c) elements that are posited (‘le p o s é ') , also
called rhematic elements (new information), and d) elements that are not overtly pre­
sent in the ‘posited’ elements but are understood or hinted at (‘le s o u s - e n t e n d u ') .
Making use of the familiar P and Q symbols, Basset argues that in many cases of ‘P
άλλά Q’, άλλά marks a break in the discourse, by correcting one of the four ele­
ments mentioned above, in as much as this is present in P, and substituting Q for P.
Q corrects, then, either a discourse theme, or a presupposed element, or a posited
element, or, finally, a s o u s - e n te n d u . By way of illustration I present here one of his
examples illustrating the correction of a presupposed element:
( R a n . 1130) Dion. ‘As-tu un reproche à faire à ces vers?’ Eur. ‘Plus de douze!’
Dion. Άλλ' ουδέ πάντα ταΰτά γ’ έστ άλλ’ ή τρία.
Here, Dionysus points out that having more than twelve points of criticism clashes
with the presupposition that there are only three lines involved. In other words,
Euripides is invited to revise the presupposition underlying his utterance
In the remainder of his paper Basset discusses various constructions of άλλά
that do not involve a ‘discursive break’, notably the frequent use after a negation, as
in ού κατ’ έπος γ ε σου κνίσω τό ρήμ’ ..., άλλά ... άπό ληκυθίου ... τούς προλό­
γους διαφθερώ (1200). He also comes back in some detail on the, rather elusive, cor­
rections of a s o u s - e n t e n d u (as in ‘He is small but strong’, the s o u s - e n t e n d u being
‘He is weak’).6 On the basis of his analyses he concludes that άλλά should be con­
sidered a corrective, rather than an adversative, conjunction.

Basset’s article is followed by that of S.R. SLINGS, who investigates adversative par­
ticles in stretches of continuous discourse, rather than in dialogue, as in Basset’s
paper. Taking Herodotus’ H i s t o r i e s as his corpus, he seeks to show, firstly, that in
such stretches, where the main story is often interrupted by embedded stories (or
sequences), these embedded sequences can be identified by means of the particles
concerned (άλλά, καίτοι and μέντοι), and, secondly, that these particles each have
an identifying function of their own. For his analysis he makes use of the hierarchical
discourse model of the Geneva school of Roulet (see above), which is supplemented
with the PUSH and POP concepts that have been developed by Polanyi and Scha.
PUSH Markers typically signal the creation of a new embedded discourse con­
stituent, while POP Markers signal a return to an embedding constituent. A third, and
more traditional, distinction used by Slings is that of relationships d e r e and d e d ic to .
Compare e.g. the adversative relationship in ‘The dress is not red but green’ ( d e r e )

6 1 should add here that this semantic phenomenon is also discussed in the papers by Slings and
Rijksbaron; unfortunately the terminology—as often—differs, Slings’ term for ‘sous-entendu’ being
‘expectation’ (cf. p. 106), and that of Rijksbaron ‘implication’ (p. 189). Much as harmonization
seems desirable, a discussion of the pro’s and con’s of these terms would fall outside the scope of
this introduction.
 INTRODUCTION

and ‘You look awful, but we’re not here to talk about your health. How’s your
work?’ ( d e d i c t o ) . In Greek, άλλά is found in the same functions as b u t here, both
within the sentence and at the level of larger stretches of discourse. To illustrate
Slings’ approach I present here one of his examples, Hdt. 4.32, a d e r e relation with
an embedded sequence marked by PUSHes:
Ύπερβορέων δέ π έ ρ ι... οντε η Σκύθαι λέγουσι... ούτε τινές άλλοι των τούτη
οίκημένων, push εί Ρύ άρα Ίσσηδόνες· ώς δ' έγώ δοκέω, ούδ’ ούτοι λέγουσι
ούδέν push έλεγον γάρ άν καί Σκύθαι, ώς περί των μουνοφθάλμων λέγουσι.
POP άλλ’ Ήσιόδω μέν έστι περί Ύπερβορέων είρημένα, έστι δέ και Όμήρω έν
Έπιγόνοισι...
Observe that άλλά replaces an element that is mentioned before the PUSH: the
embedded sequence is ignored.
Of the three particles, άλλά functions predominantly as a POP particle, καίτοι
as a PUSH particle, while μέντοι may have both functions. Both as a POP and as a
PUSH particle μέντοι is the particle p a r e x c e l l e n c e to deny an expectation (a s o u s -
e n te n d u , in Ducrot’s terminology) raised by a previous statement. As for καίτοι, this
is the ‘mirror’ particle of μέντοι. It marks a relation that Slings calls ‘inverted denial
of expectation’, aptly illustrated by him by the following pair: ‘He is rich μέντοι he
is unhappy’ as against ‘He is unhappy καίτοι he is rich’. In his conclusion he
argues that all three particles are clearly adversative.

The adversative relay is taken over by Bernard JACQUINOD, in ‘Sur le röle pragma-
tique de κα ίτο ι’. Having noted that Denniston’s classification of καίτοι seems
unduly elaborated—while at the same time acknowledging that Denniston’s observa­
tions are often excellent—he tries to achieve a simpler description by paying attention
primarily to its pragmatic function. Using Plato as his corpus he argues that καίτοι
has the following general function: it introduces a new element into an argument, an
element which does not emanate from the context, and which invites the interlocutor
to reconsider what has just been said. In actual practice it typically signals that a con­
clusion that seems to arise from some statement should, in fact, not be drawn. Thus,
at A p o l o g y 17a, οϋτω πιθανώς έλεγον might lead to the conclusion that what
Socrates’ prosecutors said was true. The ensuing καίτοι-sentence (καίτοι αληθές
γε ... ούδέν είρήκασιν) serves to prevent this conclusion. Jacquinod further argues
that the frequent use of καίτοι in (informal) syllogisms, which Denniston discusses
under a heading of its own (‘Logical’), is actually nothing but a specialized use of its
general argumentative function. In the second part of the paper καίτοι is briefly
compared with some uses of άλλά that look similar, and with the use of δέ γε and
άλλά μήν in syllogisms. As for άλλά, Jacquinod observes that while άλλά marks a
real break [cf. Bassetj in the argument, καίτοι introduces rather something contradic­
tory or at best contrastive. In an analysis of some passages where the two occur side
by side he argues that they in principle cannot be interchanged, δέ γε and άλλά μήν,
too, behave quite differently from καίτοι. Thus, άλλά μήν in no way serves to pre-
 ALBERT RIJKSBARON

vent an undesirable conclusion; it introduces rather another consideration that is rele­

vant for the argument at hand.
The paper concludes with some thoughts on the use of καίτοι in drama, which
turns out to be largely the same as that in Plato.

The next paper, ‘ή γάρ in Questions’ by A. Maria van Erp TAALMAN Kip , is the
first of two articles dealing with panicles in questions, notably in Plato. TH γάρ is
both used in elliptical questions, seeking confirmation of one’s own words, and in
questions proper; it is the latter use that is the subject of Van Erp Taalman Kip’s
paper. In real questions ή γάρ is well attested in tragedy; unlike elliptical ή γάρ, it is
used when a character asks for confirmation of something said, implied or suggested
by another speaker. The tone is urgent, and expresses various kinds of emotions.
This use is illustrated by a number of examples from all three tragedians. Having
observed that outside tragedy ή γάρ in real questions is either absent (as in Aristo­
phanes) or very rare (prose dialogue). Van Erp Taalman Kip focuses on its use in
Plato, where in the standard text editions there are only seven occurrences (as against
countless cases of elliptical ή γάρ). Four of these questions are in conformity with
their use in tragedy, since they can be shown to express a strong emotional (or mock-
emotional) tone. But the remaining three are completely different, and after a thor­
ough discussion of their context Van Erp Taalman Kip concludes that these ques­
tions are actually better taken as elliptical questions, and should be punctuated
accordingly (ή γάρ;).

With the next paper, C.MJ. SiCKiNG’s ‘Particles in Questions in Plato’, we return to
the study of argument structure in Plato (cf. Jacquinod). Sicking’s aim is to show
that a number of question particles used by Socrates serve to enable his interlocutors
to trace the articulation and overall design of the argument. Of these panicles ούν,
οΰκοΰν and (οΰκ) άρσ are the most imponant ones. Staning with οΰκοΰν, Sicking
immediately observes that in analysing this particle we are confronted with two major
problems, viz. whether we should accentuate ούκουν or ούκοΰν, and whether we
punctuate as a statement or as a question. After a discussion of some controversial
cases he submits that if ούκοΰν-questions outside Plato (and Xenophon) are rare this
is because they are bound up with Socrates’ characterististic way of asking ques­
tions. As examples he takes a number of passages from G o r g i a s , confronting
ούκουν with άρσ, δέ, πότερον ... ή, and άρσ. He argues that (1) άρσ and πότερον
... ή introduce the fundamental statements upon which the argument will be based; (2)
άρα occurs in statements that confront Socrates’ interlocutor with the consequences
of his having agreed to an earlier statement; (3) these earlier statements are introduced
by δέ; and (4) οΰκοΰν (and ούν alone) indicates that Socrates is switching from pre­
liminary or subsidiary material to the application, of that material. While άρα and dp'
οΰν-questions are veritable yes/no-questions, ούκοΰν-questions are basically assent
seeking questions. He also notes that οϋκουν-questions are completely different, for
these express disbelief with respect to the words of the interlocutor. He concludes
 INTRODUCTION

this section with some critical remarks on editorial practice concerning άρα; this is
sometimes followed by a full stop and sometimes by a question mark. Sicking advo­
cates taking άρα everywhere as a question word. In the final part of the paper a num­
ber of other question particles come to the fore, notably τί ουν and ή δέ. Sicking
argues that these particles, too, have a clear function of their own, which is summa­
rized by him as follows: τί δέ opens a new unit of development, by introducing a new
question about a new topic, while τί o w signals that the preceding question derives
its relevance from being a stepping stone for the one that is to follow.

The next contribution, Irene DE JONG’s ‘Γάρ Introducing Embedded Narratives’,

brings us back to narrative structure. She observes that Denniston, on p. 60, consid­
ers a number of instances of γάρ explanatory, although he confesses that he is unable
to establish the precise function of γάρ. Among these is a passage from Herodotus
(1.59.1), where γάρ, in fact, does not seem to have its regular function, viz. to explain
what precedes. De Jong argues, however, that if we take γάρ here not as a sentence
particle but as a particle that introduces an—embedded—narrative, there is nothing
amiss with its explanatory function. That is, it is the whole story about Pisistratus’
rise to power that constitutes an explanation of the preceding sentence. In her paper
she discusses a number of passages from various authors, ranging from Homer to
Apollonius of Rhodes, that exhibit a similar structure. De Jong offers two explana­
tions for this particular use of explanatory γάρ. The first is connected with the use of
γάρ in epic regressions. These are involved when the narrator mentions an event, and
then, from the next sentence onward, goes back in time to explain how this event
came about. The second is the use of γάρ after introductory statements of the type ‘I
am going to tell you ...’, ‘you will learn...’, as in Sophocles, O T 1240-1241: πεύση
τά κείνης άθλιας παθήματα. / όπως γάρ ... κτλ. She also observes that in all such
cases γάρ provides background information, and operates as a PUSH-particle.

In RUKSBARON’s paper, ‘Adverb or Connector? The Case of κ α ί... δέ’, the primary
focus of attention is syntax. As with all particle combinations consisting of particles
that each on their own may connect, the question arises which of the two is the con­
nective. The opinions thereon are roughly divided as follows. Denniston and Kiihner-
Gerth consider καί the connective, δέ being adverbial and meaning something like
‘on the other hand’. Most commenators, on the other hand, if they comment at all,
take δέ as the connective and καί as adverbial: ‘and also’, as in Th. 2.36.1 δίκαιον
καί πρέπον δέ. Rijksbaron argues that the latter position cannot be defended: if δέ,
in the example just mentioned, acts as a connector one would expect that it may also
connect two adjectives all by its own. This, however, is not the case (*δίκαιον
πρέπον δέ), except under special cirrcumstances. On the other hand, single καί is, of
course, the connector p a r e x c e l l e n c e between two adjectives. A further argument
against taking δέ as the connector is the existence of (καί ... ) καί ... καί ... δέ,
where the two (or more) κ α ί’s must be taken as corresponsive. Syntactically, these
sequences do not differ from e.g. κ α ί... κ α ί... δή, with adverbial δή. On the analogy
 ALBERT R1JKSBARON

of such sequences, δε, in κ α ι... δε, should be taken as an adverb, too. Its function is
a pragmatic one: it presents the second item as distinct from the first, and indicates
that that item should be considered in its own right, not as just an addition to the first
one. In English, its effect can be rendered e.g. by ‘on the other hand’, as already
advocated by Kühner-Gerth, or by ‘for that matter’. The same analysis can be
applied to another use of κ α ί... δε, viz. as a connector of clauses and sentences, as
in: καί ό πατήρ άεί λέγει καί ου φής καί οί άλλοι δέ άπαντες όμολογοΰσιν. In
such cases δέ often has the specific pragmatic function of marking a Topic shift.
There are, to be sure, a number of examples where an ‘and also’ interpretation seems
obvious, with δέ as the connector. Rijksbaron argues, however, that on closer inspec­
tion there is no need to take them this way.

The last paper with a strongly pragmatic orientation is Gerry WAKKER’s ‘Emphasis
and Affirmation: Some Aspects of μήν in Tragedy’. Her point of departure is the fact
that several particles, notably μήν, ή and δη, are called, by Denniston and others,
‘emphatic’ and ‘affirmative’, which suggests that there functions are more or less
similar. The aim of her paper is precisely to show that they are not similar, and that
their differences can be brought to light by using the pragmatic framework of Kroon
sketched above, and described in some detail by Wakker herself. Wakker argues that
μήν functions primarily at the interactional level: it signals that the speaker personally
vouches for the truth of his words, thereby reacting to something said or implied by
his addressee. This use of μήν is typically bound up with dialogues, for instance in
so-called ‘assenting’ καί μήν, e.g. after a request. The combination ή μήν is dis­
cussed by Wakker in some detail. She argues that ή, unlike μήν, functions at the rep­
resentational level; in declarative sentences it expresses the idea that the proposition is
undeniably true. *H μήν, therefore, affirms the truth of some proposition both objec­
tively, by ή, and subjectively, by μήν. In this connection she points to several distri­
butional differences between ή and μήν; thus, ή is frequently used in questions, ask­
ing whether something is really true, while μήν is hardly used in questions at all. Μήν
is also used in monologues and narratives. Here, of course, it cannot react to some­
thing said by the addressee; rather, it serves to contradict any conclusions the
addressee might draw from what the speaker himself has put forward. In the last part
of her paper, Wakker discusses ‘adversative’ and ‘progressive’ μήν. As to the for­
mer, she argues that there is no need to assign an ‘adversative’ value to the particle
itself; such adversative nuances as are present result from the contrastive nature of the
states of affairs that precede and follow μήν. As to the ‘progressive’ use, μήν may, in
fact, mark the introduction of an, usually unexpected, new element in the context, e.g.
in enumerations and when a new character appears on the stage (καί μήν ό ρώ ...).
But here, too, it has its normal attitudinal value, since it expresses that the element
concerned may be unexpected but is nevertheless real.

With the last three papers we leave the domain of syntax and pragmatics, and turn to
the history of linguistics (Sluiter) and historical linguistics (Wathelet, Waanders).
 INTRODUCTION

Ineke S lu tte r opens this series with ‘Parapleromatic Lucubrations’. She deals with
an important question that so far has remained in the background, viz. the position of
‘particles’ within an overall classification of the ‘parts of speech’.7 More specifi­
cally, she concentrates on the question of how the ancient grammarians treated the
‘conjunctions’ (σύνδεσμοι), and how they came to distinguish, within that category,
a class of σύνδεσμοι παραπληρωματικοί. The latter group consists, according to
Apollonius Dyscolus, of particles that definitely do have some kind of meaning but
that nevertheless lack a semantic and syntactic impact or purpose, and are used pre­
dominantly to ‘fill out’ or embellish metre and style. Sluiter traces the genesis of this
category from Aristotle onwards. As is well known, Aristotle considered all s y n -
d e s m o i meaningless, but after him this extreme position was given up, and there arose
a consensus that a number of them did have meaning. Those that did not—the para­
pleromatic ones—became the object of mainly stylistic theories. Thus, they were
regarded as an important means to avoid hiatus. Also, their presence or absence was
seen as a feature of the three main styles; the ‘severe’ style, for instance, should avoid
additions that had a smoothening effect, but for the ‘elegant’ style, that sought to
achieve pleasure and beauty, such additions were considered useful. Sluiter concludes
by pointing out that in the ancient classificatory system parapleromatic conjunctions
never came to be considered adverbs, the reason being that to be classified as an
adverb a word had to have meaning.

Starting from Ruijgh’s findings about κε(ν) and αν, Paul WATHELET investigates,
in ‘Les particules κε(ν) et dv dans les formules de l’épopée homérique’, the role of
these particles in the formulaic composition, with special reference to possible dialec­
tal preferences of κε(ν) and dv and to the implications of these preferences for the
composition of the epics. He observes, first of all, that there was a tendency for κε(ν)
and dv to be combined with other particles, as well as with adverbs and conjunctions,
at certain fixed, formulaic, positions in the verse. The properties of the other elements
give us certain indications concerning the relative antiquity of these combinations.
Thus, because of νυ, καί νύ κε(ν) is probably Achaean; likewise εις ο κε(ν) is
probably (Asiatic) Aeolic, because of είς. Such preferred combinations exist also
with other words, and these may, again, provide us with important clues as to their
antiquity and dialectal provenance. Thus, the combination of κε(ν) with subjunctives
in -μι, -σθα and -σι must have its origins in Asiatic Aeolic. As to the origins of κε(ν)
and dv themselves, Wathelet observes that there are very few, approximately fifty,
instances of dv that cannot be reduced to κε(ν). In principle, these irreductible cases
could stem either from the Achaean or from the Ionic stage of composition. However,
the latter is far more likely, since dv does not occur in ancient formulae. This brings
Wathelet to his last point: are there any indications that κε(ν) d o e s occur in ancient
formulae? Κε(ν) is used in a number of archaic constructions: with the subjunctive in

7 It w ill be noticed th at in the present v olum e the question: ‘W h at is a particle?’ is n o t asked. For
some literature abou t this controversial m atter I refer to the bibliographies o f Sluiter and M orpurgo
Davies.
 ALBERT RIJKSBARON

main clauses, with the future indicative, with the optative in conditional clauses. An
analysis of these constructions, however, leads to the conclusion that, although they
may be archaic, one can hardly speak of formulaic uses. From this he concludes, in
turn, that these constructions belonged to the living language of the singers until the
very end of the Aeolic stage and up to the time of Homer himself.

In the last contribution, F.M.J. WAANDERS approaches ‘particles’ from an alto­

gether different angle. The aim of his, as he himself admits, rather speculative enter­
prise is to set up an inventory of primitive particles that belonged to an archaic stage
of Proto-Indo-European, and that are at the basis of, in te r a l i a , prepositions, conjunc­
tions and particles in the regular sense, in Greek as well as in the other I-E languages.
Waanders argues that these primitive particles consisted of some consonant plus the
full grade of *e/o. He illustrates his approach by distinguishing seven particles
expressing spatial relationships: * b he l o ‘(A) with (B)’, cf. Greek αμ-φω, * p e i o
‘away and down’?, perhaps at the basis of à - π ό , h 2e ! o ‘up and away’?, cf. ά-πό,
* k e lo ‘here with me’?, cf. καί, * m e ! o ‘among’, cf. με-τά, * n e / o ‘away’, ‘down’, cf.
(έ)κεννος, perhaps also έν(ί), * h ,e lo ‘therewith’, to be found in several pronouns.

After this survey of the contents it is high time to try and draw the balance, and to
give some indications as to what a reseach programme in Greek particles might look
like, and especially, which elements of such a programme may be expected to yield
further new insights. In sketching the contours of this programme I shall concentrate
on particles in the ‘Dennistonian’ sense, for no principled reason but simply because
they are the focus of attention in the majority of the anicles collected in this book, and
lend themselves more easily to a collective appraisal.
In connection with the articles by Duhoux and Morpurgo Davies I already hinted
at what to my mind is an obvious conclusion from their investigations: instead of
treating ‘the’ particles as one monolithic block in ‘the’ Greek literature, there is a
need for studies dealing with (groups of) particles in specific dialects, genres, authors
and discourse types. In this volume, the articles by De Jong, Sicking and Van Erp
Taalman Kip demonstrate that a careful analysis of individual particles in specific
discourse types can considerably further our understanding of their function.
Again, it was noted that much remains to be done on the syntactic side (see also
below); the results of Rijksbaron’s investigation suggest that other combinations of
καί, notably the highly elusive καν (...) γάρ, might profit from the analysis of καν
(...) δε.
I already emphasized that the pragmatic framework developed by Roulet, and
elaborated by Kroon for Latin, provides us with useful analytical tools. Thus, as
Wakker’s article shows, it seems wise to try to establish at which discourse level a
given particle functions, and then to analyse particles that seemingly have the ‘same’
meaning (like ή and μην) by taking these levels into account. In this way hyper-
inflated terms like ‘emphatic’ can perhaps be given a new, more meaningful, life.
 INTRODUCTION 13

One of the concepts that turned out to be relevant for a number of authors is that
of ‘adversativity’. But it also turned out that these authors (Basset, Jacquinod, Slings)
differed in their assessment of the degree to which the particles they dealt with
(αλλά, καίτοι, μέντοι) are, in fact, adversative, rather than, e.g., corrective, particles.
So a prolonged discussion of this notion (and other central semantic notions, for that
matter, e.g. ‘additivity’, in connection with τε, καί and δέ, and ‘causality’, in con­
nection with γάρ) would not seem superfluous.
Besides the Roulet-Kroon framework, the pragmatic model developed by Ducrot
and his collaborators offers also a promising field of research, especially for the use
of particles in drama. Thus, Basset demonstrates that, by applying this model to
Greek άλλά, one can describe in much detail exactly what it is that is corrected or
contradicted by άλλά.
A striking feature of many articles (cf. Basset, Slings, Jacquinod, Sicking, De
Jong, Rijksbaron, Wakker) is that, while they describe the respective particles in
detail, they at the same time assign them one general function, mostly of a pragmatic
nature. Compare Basset’s ‘discursive break’ in connection with άλλά, Slings’
‘PUSH’ and ‘POP’-functions in connection with άλλά, καίτοι and μέντοι, and the
proposals of Jacquinod (καίτοι invites the interlocutor to reconsider what has just
been said). Sicking (αρα and άρ' οΰν-questions are veritable yes/no-questions,
ούκοΰν-questions are assent seeking questions). De Jong (γάρ in embedded narra­
tives has its regular explanatory function), Rijksbaron (δέ in κ α ι ... δέ basically
expresses distinctness) and, finally, Wakker (μην signals that the speaker vouches for
the truth of his statement). As a corollary, much weight is given to the nature of the
context to explain the various uses of a given particle. It may be expected that the
combination of detailed analysis and pragmatic ‘reductionism’ will prove rewarding
for other particles as well.8
Needless to say, the above inventory does not cover the whole field. Thus, one
important issue that was not discussed during the colloquium is that of the relation­
ship between ‘connectivity’ and asyndeton. More in particular, the point to be settled
is whether particles like γε, οΰν, δή, τοίνυν do or do not function as connectors in
the syntactic sense, i.e. in the same way as, for instance, άλλά and καί, and if not,
whether they may have, as Denniston claimed (cf. his remark quoted on p. 2), a
‘mitigating’ effect on asyndeton (not to mention the question what this ‘mitigation’
amounts to, and what is the effect of ‘unmitigated’ asyndeton). Also, some particles
were not discussed at all by Denniston, e.g. ή and νυν. (Nor were άν and κε, but
they are s u i g e n e r is ) . These lacunae remain to be filled.

In the Preface to the first edition (quoted on p. v of the second edition) Denniston
wrote: ‘These general works (viz. those of Hartung and others—A.R.) have been
supplemented during the last hundred years by a stream of dissertations, “Pro­
gramme”, and articles on individual authors; a stream which, unhappily, shows signs

8 The particle studies by Bakker, Sicking and Van Ophuijscn (for references see Sicking's article)
start from the same assumption.
 ALBERT RIJKSBARON

of drying u p ...’ No doubt Dcnniston would have been amazed at the torrential effect
on the field of the publication of his own work. There are clear signs, however, that
the field is recovering, and the renewed interest in particles is a very welcome devel­
opment. For, rather than as ‘parapleromatic’ embellishments, particles should be
regarded as part and parcel of the Greek language. In the quotation used as a motto to
this Introduction,9 Jespersen compared the function of particles and other ‘gram­
matical words’ with that of policemen controlling the traffic, a daring but appropriate
comparison. Without particles human communication would of course still be
possible, but soon look like the traffic in Cairo at rush-hour.

9 Taken from ‘Monosyllabism in English’, the Biennial Lecture on English Philology, read before
the British Academy, Nov. 6, 1928, printed in: Linguistica, Copenhagen-London 1933, 384-408;
the quotation at p. 404.
 GREC ÉCRIT ET GREC PARLÉ
Une étude contrastive des particules aux Ve-TVe siècles

Yv es du h o u x
Université de Louvain-la-Ncuve

1. Généralités

Dans son ouvrage désormais classique sur le ιτε épique’, C. J. Ruijgh a pris la peine
d’examiner en détail chacun des quelque huit cents passages homériques de cette
particule, pensant, à juste titre, qu’il est préférable d’étudier l’intégralité d’un matériel
plutôt qu’une sélection restreinte.
C’est en m’inspirant de son exemple que je voudrais présenter ici une étude met­
tant en jeu l’ensemble des particules de plusieurs oeuvres grecques des Ve-IVe s.
avant J.-C. J’y explorerai principalement la question du rapport entre grec écrit et
parlé. Ceci mettra en jeu des comparaisons entre textes dialogués ou non, ainsi
qu’entre oeuvres scéniques et non scéniques. Subsidiairement, j ’opérerai certaines
confrontations d ’auteur à auteur.
L’essentiel du corpus utilisé comportera cinq couples de textes totalisant 62.219
mots dont j ’ai opéré un dépouillement personnel; l’étude portera sur 9.958 particules
présentes dans cet ensemble. Accessoirement, je me référerai à d ’autres relevés effec­
tués par d’autres ou par moi-même.
L ’ensemble du travail présente une méthode d ’approche des particules qui pour­
rait se révéler féconde pour des recherches ultérieures.

2. Particules étudiées
Tout travail sur les particules grecques devrait, théoriquement, en avoir donné au
préalable une définition et déterminé la fonction. Ceci demanderait à soi seul une
étude à part,1 de sorte que je me limiterai ici à donner ma définition personnelle.
J'entends par p a r t i c u l e s un ensemble hétérogène de mots invariables dont la fonction
est de spécifier:

1 Sur cette question, voir par exemple F. R. Adrados, Nueva Sintaàs del Griego Antiguo. Madrid,
1992, 708-715; Blomqvist, 20-22; Denniston, xxxvii-lv; Fraenkel; Hellwig; Labéy, 1-3; J.
Redondo, ‘De nucvo sobre la cuestion de las particulas griegas’, Actas del VII Congreso Espanol
de Estudios Clâsicos, Madrid, 1989,261-266: Ruijgh, 99-102; Schwyzer, II, 553-556; Sicking.
 YVES DUHOUX

(a) une liaison entre deux éléments de même niveau hiérarchique (ainsi, καί,
‘et’)—ce sont les c o n j o n c tio n s d e c o o r d in a tio n ;
( b ) l’intonation de tout ou partie du discours (ainsi, ή, dont une des fonctions est
de souligner le caractère interrogatif de ce qui suit)—ce sont les in to n a te u r s ;
(c) le degré de réalité prêté à un élément du discours—ces particules peuvent
notamment indiquer que la portée d’un élément doit être augmentée ou diminuée
(ainsi, γε, à valeur intensive ou restrictive)—ce sont les m o d a lis a te u r s .

Comme les adverbes, les particules sont invariables, mais elles diffèrent d’eux en ceci
qu’elles ne peuvent pas être employées de façon autonome: elles doivent obliga­
toirement être utilisées avec d ’autres mots, alors qu’un adverbe peut, à lui seul,
constituer un énoncé (ainsi. Καλώς, ‘Bien’).2
Dans ce qui suit, je m’intéresserai aux particules qui ont une fréquence suffi­
samment élevée pour pouvoir faire l’objet d ’observations ne mettant pas trop le
hasard en jeu. Pour l’époque classique, le précieux dépouillement de Cauquil-
Guillaumin permet de se faire une idée relativement précise des particules les plus
importantes. Dans un corpus de ± 622.5683 mots grecs des Ve-IVe siècles, les par­
ticules les plus fréquentes sont les suivantes:4

1. καί: ± 30.000
2. δέ: ± 20.000
3. μέν: ± 10.400
4. γάρ: ± 7.200
5. ctv: ± 6.600
6 . αλλά: ± 6.400
7. ή: ±4.400
8 . τε: ± 3.300
9. ούδέ: 2.675
1 0 . γε: ± 2.200
11. οΰν:± 2.100
1 2 . ούτε: 2.018
13. τοίνυν: 1.235
14. δή: 1.214
15. μηδέ: 778
16. μήτε: 723

2 Ceci explique que je ne considère pas comme des particules les négations o\> et μή, qui peuvent
toujours constituer un énoncé à elles seules. En revanche, constituent des particules un certain
nombre de mots formés par la réunion d’une négation et d ’une particule (ούδέ/μηδέ, etc.).
3 Le signe ± indique que, en raison de lacunes dans plusieurs index utilisés, il a fallu estimer—mais
sur base statistique—le nombre d’occurrences de certains des mots en jeu. Sur la valeur de cette
estimation, voir note 5.
4 La liste que voici reprend toutes les particules dont la fréquence est supérieure à 1/10.000. Les
syntagmes constitués par juxtaposition ou combinaison de particules sont incorporés dans les occur­
rences de chacune de leurs composantes. Les seules particules de cette liste non étudiées par Dennis-
ton sont άν, αΰ, ή, νυν et όμως.
 GREC ÉCRIT ET GREC PARLÉ 17

17. μήν: 659

18. καίτοι: 576
19. όμως: 345
20. άρα: 334
21. ούκοΰν: 246
22. δήπου: 245
23. μέντοι: 240
24. ή: 222
25. τοι: 222
26. άρα: 218
27. αδ: 217
2 8 . δήτα: 188
29. νυν: 103
30. ούκουν: 90
31. γοΰν: 69

De cette liste, j ’éliminerai seulement la particule modale άν, qui est la seule de toutes
à ne pouvoir être associée qu’avec une forme verbale, et dont l’usage appartient donc
spécifiquement à la syntaxe du verbe.
Le dénombrement ci-dessus permet d ’avoir une bonne idée de l’importance
quantitative des particules en grec classique: les 31 lemmes représentent un total de ±
105.217 occurrences, soit ± 16,9 % des mots. On peut donc considérer que les textes
oratoires qui constituent l’essentiel de ce corpus comportent une particule tous les six
mots en moyenne.5 On verra toutefois plus loin que la fréquence des particules peut
varier de manière significative d’après certaines caractéristiques contextuelles.

3. Points de méthode

3.1. Grec écrit ~ grec parlé

On sait que l’Antiquité classique connaissait un rapport entre l’écrit et l'oral différent
de ce qu’il est de nos jours. Tout écrit y était normalement destiné à être p r o n o n c é .
Ceci est manifeste pour les pièces de théâtre ou les discours, mais était vrai également
pour la plupart des autres écrits (historiques, philosophiques, scientifiques, etc.). En
effet, la technique ancienne de lecture impliquait la prononciation effective du texte:
‘Pendant toute l’antiquité, et cela jusqu’au Bas-Empire, l’usage de la lecture silen-

5 II est vrai que les nombres qui viennent d’être cités reposent sur des estimations pour l'immense
majorité des particules considérées. Toutefois, les relevés effectués pour la présente étude vont exac­
tement dans le même sens: en n’incluant pas la particule άν, on a 9.958 particules sur un total de
62.219 mots, soit 16,00 %—le total correspondant de Cauquil-Guillaumin est de ± 98.617 parti­
cules sur ± 622.568 mots, soit ± 15,84 %. La différence n'est pas significative.
 YVES DUHOUX

cieuse est resté exceptionnel: on lisait soi-même à haute voix, ou, de préférence, on se
faisait lire par un serviteur.’6
Cela ayant été dit, les oeuvres antiques n’en sont pas pour autant des textes
authentiquement oraux. Bien entendu, certaines d’entre elles ont des prétentions à
reproduire le langage oral: le théâtre en est le meilleur exemple, puisqu’il était effec­
tivement joué sur scène, c’est-à-dire représenté dans une certaine mesure comme s’il
s’agissait d’événements réels (voir § 6.3). À un degré moins élevé, on mentionnera
les parties dialoguées de textes en prose qui sont censées reproduire des conversa­
tions. Dans tous ces cas, cependant, il est clair que nous avons affaire non pas à des
transcriptions fidèles de paroles réellement prononcées, mais à des créations artis­
tiques. On ne peut qualifier d ’artistiques les interrogatoires judiciaires dont certains
discours donnent le texte.7 Ils constituent certainement une mise par écrit de dia­
logues oraux. Toutefois, il semble douteux qu’ils donnent autre chose qu’une adap­
tation à l’écrit de ce qui fut réellement prononcé. D’autre part, il s’agit de textes peu
fréquents et très brefs.
Nous ne disposons donc d ’aucune transcription scrupuleuse de textes oraux
antiques. Cette difficulté n’est pas la seule à laquelle se heurte l’étude du grec ancien
oral. En effet, les études contemporaines, où l’on dispose d ’un matériel oral et écrit
virtuellement illimité, montrent que, en tout cas dans les langues occidentales, il est
trop simple de considérer comme deux blocs monolithiques l’oral et l’écrit. Chacune
de ces deux catégories comprend, en fait, plusieurs composantes parfois très dif­
férentes les unes des autres—comparer, par exemple, une lettre personnelle ~ un
rapport écrit; un dialogue familier ~ un discours officiel ...8 II y a donc lieu de cibler
très exactement le type de discours oral ou écrit qui est visé. Dans le cas du grec
ancien, à corpus oral inexistant et à corpus écrit strictement fini, le choix est extrême­
ment limité.
Compte tenu de ces limites, la présente recherche partira de l’attique tel qu’il
ressort de textes écrits littéraires des Ve-IVe s. Il est raisonnable de penser que ces
oeuvres représentent diverses formes du grec que connaissait le public cultivé de la
classe moyenne. À travers ce type de textes, je chercherai à m’approcher de la langue
effectivement pariée par ce même public dans ses conversations ordinaires. Par con­
vention, c’est ce type bien particulier d ’attique qui sera désigné ici par ‘grec oral’ ou
‘parlé’.

6 H.-I. Manou, Histoire de l’éducation dans l’antiquité. Paris, 1965^, 234. La lecture silencieuse
existait à Athènes aux Ve-IVe s., en tout cas pour les documents épistolaires ou analogues: voir B.
M. W. Knox, ‘Silent Reading in Antiquity’, Greek Roman and Byzantine Studies 9 (1968), 432-
435.
7 Ainsi, Lysias, Contre Ératosthéne (discours XII), § 25.
8 Voir par exemple pour l ’anglais D. Biber, Variation Across Speech and Writing. Cambridge,
1988.
 GREC ÉCRIT ET GREC PARLÉ 19

3.2. Grec parlé et particules

Pour qui veut tenter d’appréhender le grec ancien parlé à travers le filtre de ses écrits,
les particules sont spécialement intéressantes et l’on a souvent spéculé sur leurs rap­
ports avec l’oral.
Une position extrême est représentée par Labéy, pour qui le grec parlé ignorait
totalement les particules. Selon lui, elles ‘appartiennent à la langue écrite. Elles
expriment les intonations de la voix. Indispensables dans les écrits, où leur absence
produirait le désordre, elles disparaissent dans la langue parlée.’9 Cette thèse est, de
toute évidence, inadmissible, entre autres parce que le grec parlé contemporain, vivant,
utilise des particules héritées.
L’avis général est plus nuancé: de même que Labéy, des auteurs comme
Blomqvist, Fraenkel ou Hellwig postulent que toutes les particules (ou que, au mini­
mum, bon nombre d ’entre elles) ont la même fonction que l ’intonation dans les
langues modernes.10 Mais en même temps, on pense souvent que les particules
étaient spécialement fréquentes dans le langage oral. Ainsi, E. Schwyzer parle de par­
ticules qui ‘charakteristisch sind für die Umgangssprache, besonders für das
Gespräch.’11 De même, Hellwig évoque ‘die Fülle der Partikeln im Griechischen, vor
allem bei der Wiedergabe gesprochener Sprache.’12 Dans un ouvrage qui fait tou­
jours autorité, Denniston est d’avis que c’est le dialogue qui reflète le mieux ce lan­
gage parlé: ‘it cannot be doubted that Greek conversation was full of particles: at
moments of excitement and emotional tension the dialogue of tragedy and comedy
fairly bristles with them.’13 Il en résulte que, pour lui, ‘the main line of cleavage is
not so much between poetry and prose as between dialogue and continuous speech or
formal exposition.’ Ce jugement d’ensemble se trouve appuyé par quelques con­
sidérations particulières. Ainsi, Denniston signale pour ή que ‘in prose, except in
certain combinations (...) the particle is almost confined to dialogue.’14 Il caractérise
τοι comme ‘exceedingly common in dialogue, prose and verse’15 et τοίνυν comme
‘much commoner in dialogue than in continuous speech.’16

9 Labéy, 3-4. De même, voir J. Humbert, Syntaxe grecque, Paris, I9603, 368: ‘Les Anciens
écrivaient de façon continue, sans séparer les mots, sans faire intervenir, à Γintérieur de la phrase ou
entre les phrases, ce que nous appelons des signes de ponctuation... Il suffit d’avoir vu l ’aspect de
continuité ininterrompue que présentent également les inscriptions, les papyrus et les manuscrits...
pour prendre une première idée du rôle indispensable que jouent ces particules qu’on appelle, d ’une
façon bien sommaire, «particules de liaison».’ Voir déjà en ce sens Meillet, 235.
10 Déjà au XVIIIe s., H. Hoogeveen pensait qu’une partie des particules se situait dans la même
sphère que le ‘vultus, gestus, vocis ductus’ propre au grec parlé (H. Hoogeveen, Doctrina panicu­
larum linguae Graecae, Leyde, 1769, cité par C. M. J. Sicking, ‘Devices for Text Articulation in
Lysias I and ΧΙΓ, dans Sicking-Van Ophuijsen, 5).
Schwyzer, II, 556.
12 Hellwig, 163.
13 Denniston, Ixxii-lxxiii.
14 Denniston, 279-280.
15 Denniston, 539.
16 Denniston, 569; repris par J. M. Van Ophuijsen, Ό ΤΝ , APA, ΔΗ, TOINTN: The Linguistic
Articulation of Arguments in Plato’s Phaedo’, dans Sicking-Van Ophuijsen, 152.
 YVES DUHOUX

Il n’y a cependant pas unanimité sur la prédilection que le langage parlé aurait
pour les particules: A. Meillet est d’avis que ‘la langue parlée courante ne faisait pas
des particules un emploi aussi constant [que le grec littéraire]’17—il est vrai qu’il
ajoute plus loin, à propos de l’oeuvre de Platon, qu’elle donne ‘une idée exacte de la
langue courante des milieux cultivés de l’époque de l’auteur.’18 P. Chantraine pense,
de son côté, qu’‘il est probable que la langue populaire ne faisait des particules qu’un
médiocre usage.’19
Il m’a paru intéressant de voir ce qu’il en était réellement de cette question et de
mettre à l’épreuve l’opposition centrale que détecte Denniston entre textes dialogués
ou non. J ’ai donc confronté l’emploi des particules dans les parties dialoguées et non
dialoguées d’un certain nombre d ’oeuvres des Ve-IVe s. Le but poursuivi est de
déterminer s’il existe des différences perceptibles entre ces deux types de contextes,
et lesquelles. En cas de réponse positive, on conclura que les caractéristiques propres
aux dialogues doivent refléter des usages de la langue parlée, puisque les passages
dialogués tendent à donner l’impression d’une véritable conversation.

3.3. Procédure mise en oeuvre

Le point de départ de l’examen a été de dénombrer les occurrences de toutes les par­
ticules retenues (§ 2) et de les mettre en rapport avec la longueur des textes où elles
figurent. Cette approche n’est pas nouvelle, puisque bien des travaux ont donné
d ’utiles relevés de fréquences de particules: fréquences absolues chez un auteur ou
dans une oeuvre donnée, mais aussi fréquences relatives (par rapport à une unité de
référence—le plus souvent la page; parfois la ligne, la proposition ou le vers). Ici, j'ai
systématiquement utilisé les ressources de l’informatique, qui permettent un
comptage précis non seulement du nombre des particules, mais aussi—et c’est là que
réside leur véritable originalité—du nombre de m o t s du corpus où elles figurent.
Cette façon de faire permet des comparaisons considérablement plus exactes et
révélatrices que la plupart des relevés antérieurs, qui ne pouvaient établir la fréquence
relative des particules que de façon nécessairement approximative.
L ’appréciation du caractère significatif des différences de fréquence observées a
été réalisée grâce au test statistique du chi carré (χ2).20 Son intérêt est de fournir un
outil parfaitement au point et totalement objectif, permettant de juger si deux
fréquences sont, ou pas, significativement différentes. Ce test compare les effectifs

17 Meiliet, 235.
18 M eillet, 241.
19 Dans Labéy, VII.
20 Sur ce test, voir par exemple Ch. Muller, Initiation aux méthodes de la statistique linguistique.
Paris, 1973, 116-127; A. Woods et alii. Statistics in Language Studies. Cambridge, 1986,132-153.
II est important de savoir que la portée des résultats obtenus varie d’après que le χ2 délecte, ou non,
une différence significative. Il n’y a démonstration probante que dans le premier cas. Si aucune
divergence n'est mise en évidence, on a simplement affaire à une absence de preuve de différence. En
ce cas.il peut n’y avoir réellement aucune différence, mais il pourrait aussi en exister une détectable
seulement dans un autre examen (par exemple en cas d’augmentation de la taille de l’échantillon).
 GREC ÉCRIT ET GREC PARLÉ 21

observés21 aux effectifs théoriquement attendus dans l’hypothèse où la répartition

des diverses variables serait identique dans chacun des échantillons. Les effectifs
théoriques du χ 2 neutralisent les différences éventuelles de proportions dans les
populations observées, de manière à répartir les variables conformément au seul jeu
du hasard.22
La fréquence des particules dans les oeuvres examinées personnellement a été
calculée sur base du texte du T h e s a u r u s L i n g u a e G r a e c a e de l’Université de Cali­
fornie à Irvine. Les citations d’autres auteurs, restitutions et athétèses ont été élimi­
nées de chaque corpus.
La recherche des particules et tous les dénombrements ont été effectués avec
l’aide du logiciel L e x i s ,23 Les syntagmes de particules (ainsi, τε καί) ne sont pas
signalés tels quels, mais sont conventionnellement incorporés dans les occurrences de
chacune de leurs composantes (dans l’exemple choisi, sous te et sous καί). En ce
qui concerne καί, je n ’ai pas effectué le tri entre ses emplois conjonctifs et adver­
biaux (au sens de ‘aussi’ ou ‘même’) et les ai indistinctement regroupés. Je n’ai pas
tenu compte des exemples de καί dans καλοκάγαθία.
Les résultats des examens seront chaque fois présentés dans un tableau. Sur les
sigles utilisés, voir § 9. Les données relatives à une particule dont la différence
d ’emploi est significative sont imprimées en grasses.
Après chaque tableau, figurent les résultats des comparaisons suivantes:
a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o ts d a n s le s te x t e s d ia l o g u é ~ n o n
d ia lo g u é
b) N o m b r e d e p a r t i c u l e s d iffé r e n te s d a n s le s te x te s d i a l o g u é ~ n o n d ia lo g u é
c) P a r t i c u l e s n e d if f é r a n t p a s d e f r é q u e n c e d a n s le s te x te s d ia lo g u é ~ n o n d i a ­
lo g u é
d) P a r tic u le s d if f é r a n t d e f r é q u e n c e d a n s le s te x te s d ia lo g u é ~ n o n d ia lo g u é .

4. Particules et dialogues

4.1. Platon, M é n o n ~ A p o l o g i e d e S o c r a te

Pour étudier les différences d ’emploi d’après le caractère dialogué ou non du texte,
Platon (± 429-347) constituait une référence obligée. J’ai donc commencé par con­
fronter l’emploi des particules dans deux de ses oeuvres de jeunesse: d’un côté, le
M é n o n , qui est presque entièrement dialogué; d’autre part, V A p o lo g ie d e S o c r a te , qui
ne comporte que quelques rares dialogues.24

21 Pas leurs pourcentages!

22 Conformément aux usages courants, j ’ai retenu 5 % comme pourcentage en dessous duquel les
différences entre des distributions sont considérées comme statistiquement significatives; les effec­
tifs théoriques doivent égaler ou dépasser 5.
22 Ce logiciel a été conçu et réalisé par Richard Goulet (C.N.R.S.). Voir à ce sujet Y. Duhoux, ‘Un
indcxeur/concordanceur automatique de textes grecs et latins: Lexis’, Syntaktika 11 (1996) 1-17.
24 Éditions utilisées par le TLG: J. Bumct, Oxford, 1903 [réimpr. 1968], 1900 (réimpr. 1967].
 YVES DUHOUX

OCCURRENCES OCCURRENCES LA DIFFÉRENCE SENS DE LA

ΕΝ TEXTE DIA- EN TEXTE NON EST-ELLE DIFFÉRENCE
LOGUÉ DIALOGUÉ SIGNIFICATIVE? ÉVENTUELLE
αλ λά 113 92 non
άρα 32 9 oui d > nd
άρα 21 3 oui d > nd
αΰ 8 11 non
γάρ 81 124 oui d < nd
γε 115 43 oui d > nd
70Û V 7 3 NC
δέ 194 177 non
δη 65 59 non
δηπου 11 8 non
δητα 7 2 NC
ή 110 86 non
ή 4 3 NC
καν 452 477 oui d < nd
κ α ίτο ι 4 7 non
UÉV 114 106 non
ιιέντοι 6 13 non
ιιηδέ 7 8 non
μιίν 10 1 oui d > nd
αήτε 16 20 non
νυ ν 0 1 NC
ÖLKOC 5 5 NC
ο υ δέ 33 34 non
ούκουν 1 1 NC
ούκούν 42 2 oui d > nd
ούν 65 68 non
ούτε 15 41 oui d < nd
τε 63 56 non
τοι 1 3 NC
τοίνυν 13 2 oui d > nd

1. Emploi des particules dans Platon, M é n o n ~ A p o lo g ie d e S o c r a te

a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o t s d a n s l e s t e x t e s d ia l o g u é ~ n o n
d ia lo g u é : 1.615 - 7.963 et 1.465 - 7.256—différence non significative.
 GREC ÉCRIT ET GREC PARLÉ 23

b) N o m b r e d e p a r t i c u l e s d i f f é r e n t e s d a n s le s te x te s d ia l o g u é - n o n d ia lo g u é : 29
~ 30—différence non significative.
c) P a r t i c u l e s n e d i f f é r a n t p a s d e f r é q u e n c e d a n s l e s t e x t e s d i a l o g u é ~ n o n
d i a l o g u é : 14/30 (άλλά, αΰ, δέ, δή, δήπου, ή, καίτοι, μέν, μέντοι, μηδέ, μήτε,
ούδέ, ουν, τε).
d) P a r t i c u l e s d i f f é r a n t d e f r é q u e n c e d a n s le s te x t e s d i a l o g u é ~ n o n d i a l o g u é :
9/30 (αρα, άρα, γάρ, γε, καί, μήν, ούκοΰν, ούτε, τοίνυν). Elles s’opposent comme
suit:
Fréquence plus élevée dans les dialogues: αρα, άρα, γε, μήν, ούκοΰν, τοίνυν.
Fréquence plus élevée dans les parties non dialoguées: γάρ, καί, ούτε.

4.2. Platon, P r o t a g o r a s

L’examen qui précède démontre que ni la fréquence totale des particules ni leur
éventail ne diffèrent significativement d ’après le caractère dialogué ou non des
oeuvres de Platon où elles figurent Toutefois, il existe des différences significatives
de fréquences pour une minorité de particules (9/30).
Bien que les deux oeuvres choisies pour l’examen précédent soient généralement
considérées comme relativement contemporaines, on doit toutefois se demander si
l’emploi qu’elles font des particules ne pourrait pas être lié à des différences
chronologiques ou autres. Pour vérifier cette hypothèse, j ’ai examiné les parties dia­
loguées ou non d ’un seul et même dialogue platonicien, le P r o t a g o r a s 25 (ce texte
pourrait être antérieur au M é n o n et à Y A p o lo g ie ) .
La répartition de ce texte en parties dialoguées ~ non dialoguées ne pose pas de
problème dans bon nombre de cas. Il arrive toutefois régulièrement que l’on doive
hésiter. J’ai alors tranché de mon mieux, considérant les passages comme non dia-
logués dès qu’il me semblait qu’était défait le lien personnel qui unit deux interlocu­
teurs dans un dialogue—ceci se produit le plus souvent en cas d ’intervention d ’une
certaine longueur. Dans les parties dialoguées, j ’ai laissé les transitions utilisées par
le narrateur pour relier deux interventions, en considérant conventionnellement
qu’elles faisaient partie intégrante du passage.
On trouvera ci-dessous les résultats de cette comparaison.

OCCURRENCES OCCURRENCES LA DIFFÉRENCE SENS DE LA

EN TEXTE DIA- EN TEXTE NON EST-ELLE DIFFÉRENCE
LOGUÉ DIALOGUÉ SIGNIFICATIVE? ÉVENTUELLE
άλλά 71 83 non
αρα 17 7 oui d > nd
αοα 24 2 oui d > nd
αΰ 12 14 non
JCSfi________ 63 111 non

25 Édition utilisée par le TLG: J. Bumet, Oxford, 1903 [réimpr. 1968].

 YVES DUHOUX

γε 54 38 out d > nd
γοϋν 2 3 NC
δέ 176 251 non
δή 42 35 oui d > nd
δππου 5 2 NC
δητα 2 0 NC
η 85 85 oui d > nd
ή 8 2 NC
καί 371 655 oui d < nd
καίτοι 1 2 NC
μέν 66 132 oui d < nd
μέντοι 14 5 oui d > nd
μτιδέ 1 5 NC
μην 5 0 NC
μήτε 0 7 NC
νυν
δμως 2 3 NC
ουδέ 22 25 non
ούκοΰν
ούκοΰν 27 0 oui d > nd
ουν 74 68 oui d > nd
οΰτε 21 24 non
τε 71 112 non
τοι 5 3 NC
τοίνυν 5 4 NC

2. Emploi des particules dans Platon, P r o ta g o r a s , parties dialoguées ~ non

dialoguées

a)
E n s e m b l e d e s p a r t i c u l e s - le r e s te d e s m o t s d a n s l e s te x t e s d ia l o g u é - n o n
1.246-6.073 et 1.678-8.345—difference non significative.
d ia lo g u é :
b) N o m b r e d e p a r t i c u l e s d iffé r e n te s d a n s le s te x te s d ia lo g u é - n o n d ia lo g u é : 27
- 25: différence non significative.
Dans ce qui suit, des caractères gras signaleront des particules d ’emploi identique
dans le M é n o n et 1’A p o l o g i e d e S o c r a te , d’une part, et, d’autre part, dans !e P r o ta g o ­
ras.
c) P a r t i c u l e s n e d i f f é r a n t p a s d e f r é q u e n c e d a n s l e s t e x t e s d i a l o g u é ~ n o n
d i a l o g u é : 7/28 (αλλά, ου, γάρ, δε, ουδέ, οΰτε, τε).
d) P a r t i c u l e s d i f f é r a n t d e f r é q u e n c e d a n s le s te x te s d i a l o g u é - n o n d ia lo g u é :
10/28 (ορα, άρα, γε, δή, ή, κα ί, μέν, μέντοι, ούκοΰν, οΰν). Elles s’opposent
comme suit:
 GREC ÉCRIT ET GREC PARLÉ 25

Fréquence plus élevée dans les dialogues: άρα, ά ρα, γε, δή, ή, μεντοι,
ούκούν, ουν
Fréquence plus élevée dans les parties non dialoguées: κ α ί , μεν.
Π existe de larges points c o m m u n s e n t r e le P r o t a g o r a s , d ’une part, et, d ’autre
part, le M é n o n et Y A p o l o g i e : ceci montre que l’emploi des particules des deux
derniers textes n’a probablement été que peu influencé par des différences
chronologiques ou autres.

4 .3 . X é n o p h o n , Banquet

Les examens antérieurs portaient tous deux sur des oeuvres de Platon. On doit donc
se demander si les résultats obtenus ne tiendraient pas à une particularité de cet
auteur. Pour vérifier cette hypothèse, j ’ai examiné les parties dialoguées ou non d’une
oeuvre de Xénophon (± 428/7-± 354), le B a n q u e t é On en trouvera les résultats ci-
dessous.

OCCURRENCES OCCURRENCES LA DIFFÉRENCE SENS DE LA

EN TEXTE DIA­ EN TEXTE NON EST-ELLE DIFFÉRENCE
LOGUÉ DIALOGUÉ SIGNIFICATIVE? ÉVENTUELLE
άλλα 50 39 non
cioct 4 0 NC
aoa 4 1 NC
av 9 2 oui d > nd
51 54 non
γε 40 35 non
γοΰν 3 7 NC
fié 95 192 oui d < nd
32 19 non
δήπου 5 4 NC
δίίτα 1 0 NC
ή 35 35 non
ri 6 1 NC
καί 230 314 oui d < nd
κ α ίτοι 0 2 NC
ιιέν 76 119 oui d < nd
ιιέντοι 13 7 non
μηδέ 5 2 NC
8 25 oui d < nd
ατπε 0 2 NC
νυν 1 0 NC

Édition utilisée par le T LG : E. C. Marchant, Oxford, 1921 [réimpr. 1971].

26 YVES DUHOUX

oucoc 3 3 NC
ούδέ 8 14 non
οΰκοΰν 1 0 NC
οΰκοΰν 13 2 ouì d > nd
οΰν 30 19 non
ούτε 4 6 NC
τε 20 56 ouì d < nd
τοι 2 3 NC
τοίνυν 10 4 non

3. Emploi des particules dans Xénophon, B a n q u e t, parties dialoguées - non

dialoguées

a) E n s e m b le d e s p a r tic u le s ~ le r e s t e d e s m o t s d a n s le s t e x t e s d i a l o g u é - non
d ia l o g u é : 759 ~ 3.875 et 967 ~ 3.823—différence significative. Le pourcentage des
particules est le suivant dans les textes dialogués ~ non dialogues: 16,38 % ~ 20,19
%. Il y a davantage de particules dans les parties non dialoguées.
b) N o m b r e d e p a r t i c u l e s d i f f é r e n t e s d a n s l e s te x t e s d i a l o g u é ~ n o n d i a l o g u é : 28 -
26: différence non significative.
c) P a r t i c u l e s n e d if f é r a n t p a s d e fr é q u e n c e d a n s l e s te x te s d ia lo g u é ~ n o n d ia lo g u é :
9/30 (αλλά, γάρ, γε, δή, ή, μέντοι, ούδέ, οΰν, τοίνυν).
d) P a r t i c u l e s d if f é r a n t d e f r é q u e n c e d a n s l e s te x t e s d ia lo g u é ~ n o n d ia l o g u é : 7/30
(αΰ, δέ, καί, μέν, μην, οΰκοΰν, τε). Elles s’opposent comme suit:
Fréquence plus élevée dans les dialogues: αΰ, οΰκοΰν.
Fréquence plus élevée dans les parties non dialoguées: δέ, καί, μέν, μήν, τε.
Le B a n q u e t a un usage en panie nettement different de celui de Platon. Le point cru­
cial est une fréquence de l’ensemble des particules plus grande dans les parties non
dialoguées que dans les autres.

4.4. Aristophane, T h e s m o p h o r i e s

Xénophon s’oppose à Platon par l'empioi plus frequent des particules dans les par­
ties non dialoguées qu’ailleurs. Il faut se demander si cet usage lui est propre ou non.
Pour répondre à cette question, j ’ai examiné les parties dialoguées ou non d’une
pièce de théatre, les T h e s m o p h o r i e s d’Aristophane27 (datées de 411). Cette oeuvre
offre l ’avantage d’etre écrite en vers, ce qui l’oppose aux textes choisis jusqu’ici, qui
étaient tous en prose.
 GREC ÉCRIT ET GREC PARLÉ 27

OCCURRENCES OCCURRENCES LA DIFFÉRENCE SENS DE LA

ΕΝ TEXTE DIA- EN TEXTE NON EST-ELLE DIFFÉRENCE
LOGUÉ DIALOGUÉ SIGNIFICATIVE? ÉVENTUELLE
αλλά 45 24 non
άοα 10 1 NC
αοα 4 0 NC
αυ 9 2 NC
Υάρ 54 30 non
γε 51 7 oui d > nd
γοΰν 1 2 NC
δέ 59 68 oui d < nd
7 9 non
δήπου 0 2 NC
δήτα 15 0 oui d > nd
ΐ» 9 22 oui d < nd
2 0 NC
καί 111 120 oui d < nd
κα ίτοι 0 1 NC
uév 19 24 non
μέντοι 5 1 non
μηδέ
μην 5 0 NC
μήτε 4 2 NC
νυν 11 3 non
οαως 0 2 NC
ούδέ 5 14 oui d < nd
οϋκουν 1 0 NC
ούκουν 1 0 NC
οΰν 24 9 non
ούτε 4 0 NC
te 19 39 oui d < nd
τοι 7 4 NC
τοίνυν 7 2 NC

4. Emploi des particules dans Aristophane, T h e s m o p k o r i e s , parties dialoguées ~

non dialoguées

a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o t s d a n s le s t e x t e s d i a l o g u é ~ n o n
d ia lo g u é : 489 - 3.475 et 388 ~ 2.424— différence non significative.
28 YVES DUHOUX

b) N o m b r e d e p a r t i c u i e s d i f f é r e n t e s d a n s l e s te x te s d i a l o g u é ~ n o n d ia lo g u é : 26 ~
22: différence non significative.
c) P a r t i c u i e s n e d i f f é r a n t p a s d e f r é q u e n c e d a n s l e s te x te s d ia l o g u é ~ n o n d ia l o g u é :
6/29 (άλλά, γάρ, δή, μέν, νυν, οΰν).
d) P a r t i c u i e s d i f f é r a n t d e f r é q u e n c e d a n s l e s te x t e s d ia l o g u é ~ n o n d i a l o g u é : 7/29
(γε, δε, δήτα, ή, καί, ουδέ, τε). Elles s’opposent comme suit:
Fréquence plus élevée dans les dialogues: γε, δήτα.
Fréquence plus élevée dans les parties non dialoguées: δε, ή, καί, ουδέ, τε.
Dans les T h e s m o p h o r ie s , Aristophane a des usages relativement proches de ceux de
Platon: ni la fréquence totale des particuies ni leur éventail ne different significative-
ment d’après le caractère dialogué ou non des passages. De mème, on observe des
différences significatives de fréquences pour une minorité de particuies prises indi-
viduellement. On peut done conclure que l ’usage de Xénophon (§ 4.3) lui est
apparemment propre.

4.5. Sophocle, O e d ip e à C o io n e

L’examen précédent vient de nous montrer les conditions d’emploi des panicules
d ’après le caractère dialogué ou non du contexte chez Aristophane. On peut cepen-
dant se demander si, dans les oeuvres versifiées, les résultats obtenus ne seraient pas
limités aux comédies.
Pour vérifier cette hypothèse, j ’ai examiné les parties dialoguées ou non d’une
oeuvre tragique, Y O e d i p e à C o io n e de Sophocle28 (représenté en 401). On en trou-
vera les résultats ci-dessous.

OCCURRENCES OCCURRENCES LA DIFFERENCE SENS DE LA

ΕΝ TEXTE DIA- EN TEXTE NON EST-ELLE DIFFÉRENCE
LOCUÉ DIALOGUÉ SIGNIFICATIVE? Ev e n t u e l l e
άλλά 51 41 non
άοα 5 2 NC
άοα 10 5 non
αΰ 2 4 NC
γάρ 88 57 non
γε 49 18 ouì d > nd
γοΰν 2 0 NC
δέ 132 126 non
δή 9 7 non
δήπου
θήτα 13 3 oui d > nd
ή 29 24 non
Λ ____________ 16 1 oui d > nd
 καί 155 142 non
καίτοι 1 3 NC
μέν 35 47 ouì d < nd
μέντοι 0 3 NC
ατιδέ 11 7 non
7 2 NC
5 10 non
νυν 5 3 NC
4 3 NC
ούδέ 18 28 oui d < nd
οϋκουν 2 2 NC
ούκοΰν
ούν 10 4 non
ούτε 8 22 oui d < nd
τε 45 58 oui d < nd
τον 9 7 non
τοίνυν 1 0 NC

5. Emploi des particules dans Sophocle, O e d ip e à C o io n e , parties dialoguées ~

non dialoguées

a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o t s d a n s l e s t e x t e s d i a l o g u é ~ n o n
d ia lo g u é : 722 ~ 4.834 et 629 ~ 4.193—différence non significative.
b) N o m b r e d e p a r t i c u l e s d i f f e r e n t e s d a n s le s te x te s d i a l o g u é ~ n o n d ia l o g u é : 27 ~
26: différence non significative.
c) P a r t i c u l e s n e d i f f é r a n t p a s d e f r é q u e n c e d a n s le s te x te s d i a l o g u é ~ n o n d ia lo g u é :
11/28 (άλλα, άρα, γάρ, δε, δή, ή, καί, μηδέ, μήτε, ούν, tot).
d) P a r t i c u l e s d i f f é r a n t d e fr é q u e n c e d a n s l e s te x te s d i a l o g u é ~ n o n d ia l o g u é : 7/28
(γε, δήτα, ή, μέν, ούδέ, ούτε, τε). Elles s’opposent comme suit:
Fréquence plus élevée dans les dialogues: γε, δήτα, ή.
Fréquence plus élevée dans les parties non dialoguées: μέν, ούδέ, ούτε, τε.
L 'O e d i p e à C o i o n e de Sophocle a des usages comparables à ceux des T h e s -
m o p h o r i e s d’Aristophane. Il n’y a de différence significative d’après le caractère
dialogué ou non des passages ni dans la fréquence totale des particules ni dans leur
éventail. Par ailleurs, on observe des différences significatives de frequences pour une
minorité de particules prises individuellement. Ces caractéristiques s’observent done
dans les vers tant de comédie que de tragèdie, de mème que dans la prose de Platon.
 YVES DUHOUX

4.6. C onclusion

Le caractère dialogué ou non des textes a les conséquences suivantes sur l’emploi des
particules (les remarques ci-dessous ne mettent enjeu que les cas où le χ2 est calcu­
lable).

4.6.1. Il n’a aucune influence perceptible sur l’éventail des particules.29

4.6.2. Les parties dialoguées n’ont pas davantage de particules que les autres—et
Xénophon présente même le phénomène inverse, avec m o i n s de particules dans les
dialogues qu’ailleurs. Ceci va à l’encontre des avis donnés plus haut (§ 3.2) sur la
prédilection qu’auraient les particules pour les parties dialoguées.

4.6.3. Cette symétrie apparente dissimule toutefois une réalité infiniment plus com­
plexe. En effet, alors que la moitié des emplois des particules sont indifférents au
caractère dialogué ou non du contexte (47/87), l’autre moitié y est, elle, sensible
(40/87). Et dans cette dernière, il existe des tendances opposées: 21 emplois sont plus
fréquents dans les dialogues, contre 19 ailleurs. L’uniformité des fréquences globales
résulte donc de l’équilibre entre ces mouvements divers.
La complexité des processus en jeu est d ’ailleurs encore plus grande qu’il
semble, car le profil de chaque particule est loin d’être toujours régulier. La moitié
d’entre elles peuvent être utilisées de façon différenciée e t indifférenciée (15/30). Il y
a plus: lorsque des particules sont différenciées, elles le sont presque toujours dans le
même sens, mais il en existe deux où la différenciation s’opère en sens opposés: ή et
μην. Aucune particule n’est d’ailleurs différenciée dans tous ses emplois;30 et il n’y
en a qu’une seule qui soit toujours indifférenciée: il s’agit ά’άλλά, qui semble donc
la moins sensible au caractère dialogué ou non du contexte.31
Ces irrégularités ne tiennent visiblement pas à des différences entre auteurs,
puisqu’elles s’observent dans les oeuvres de Platon examinées: 7 particules sur les
17 où le χ2 est calculable y ont des emplois variables (γάρ, δή, ή, μεν, μέντοι, ούν,
ούτε).32
Une partie importante des particules a donc un comportement étonnamment
changeant. On en retire le sentiment que le c h o i x d e la f r é q u e n c e d e b o n n o m b r e
d 'e n t r e e l l e s p o u r r a i t , e n d é f i n i t i v e , ê tr e l a r g e m e n t a r b i t r a i r e : il ne tiendrait pas
nécessairement à leur fonction, mais au désir de l’auteur de les utiliser pour carac­
tériser leurs contextes.29301

29 Rappelons toutefois que les syntagmes de particules n’ont pas été pris en compte en tant que tels
dans notre étude (§ 3.3).
30 Mais il y en a deux qui ont des emplois différenciés dans quatre comparaisons sur cinq: γ ε et
καί.
31 Ceci ne signifie évidemment pas qu’elle ne pourrait pas différer dans d’autres échantillons.
Signalons que trois particules ont des emplois indifférenciés dans quatre comparaisons (mais pas
dans la cinquième!): γάρ, δή et οΰν.
32 On vena que ces divergences sont du même ordre que celles qui s’observent d’auteur à auteur (§
 GREC ÉCRIT ET GREC PARLÉ 31

4.6.4. Bien que largement arbitraires, les fréquences des particules sont employées
dans un but précis. Dans chaque oeuvre, elles servent à constituer un noyau restreint
de particules (de sept à dix) opposant les parties dialoguées aux autres. Leur liste est
la suivante (entre parenthèses: le nombre d ’emplois différenciés):

Particules plus fréquentes dans les parties d ia l o g u é e s : ctpa (2 ex.), âpa (2 ex.),
αΰ (1 ex.), γε (4 ex.), δή (1 ex.), δήτα (2 ex.), ή (1 ex.), μέντοι (1 ex.),
ούκοΰν (3 ex.), οΰν (1 ex.), τοίνυν (1 ex.).

Particules plus fréquentes dans les parties n o n d i a l o g u é e s : γάρ (1 ex.), δέ (2

ex.), καί (4 ex.), μέν (3 ex.), ουδέ (2 ex.), ούτε (2 ex.), τε (3 ex.).

Particules plus fréquentes dans les parties d ia lo g u é e s e t n o n d i a l o g u é e s : ή, μήν

(chaque fois 1 ex.).

En ce qui concerne les usages décrits par Denniston comme propres aux textes
dialogués (§ 3.2), les résultats de notre étude sont les suivants:
- pour ή, la forte préférence pour le dialogue attribuée exclusivement à la prose n’est
pas vérifiable à cause du trop petit nombre d ’exemples; en revanche, le phénomène
est bel et bien attesté en vers (O e d ip e à C o lo n e ) .
- pour τοι, il semble inexact que la panicule soit spécialement utilisée dans les
dialogues. En effet, la fréquence de τοι est si faible que la différence entre textes
dialogués ou non n’est pas évaluable par le χ2 dans quatre comparaisons sur cinq. Et
dans la cinquième, elle n’est pas significative.
- pour τοίνυν, qui serait de loin plus frequente dans les dialogues, le jugement semble
trop absolu. Dans une des deux comparaisons appréciables par le χ2, τοίνυν est
effectivement plus fréquente dans les dialogues, mais dans l’autre, sa différence
d ’emploi n’est pas significative.

5. Particules et textes scéniques

5.1. Les particules dans les textes scéniques ~ non scéniques

On vient de voir que la fréquence des particules contribue à opposer les textes
dialogués ou non. Mais un autre facteur ne pourrait-il pas avoir aussi influencé leur
emploi, à savoir le caractère scénique ou non des oeuvres? Il est tentant de répondre
positivement à cette question, puisque le théâtre a précisément pour ambition de
restituer la vie réelle dans une mesure supérieure à celle des autres oeuvres (voir §
6.3). Sans aller aussi loin que Denniston, pour qui ‘Greek drama reproduces, as far
as one can tell, the free use of particles in everyday speech,’33 on pourrait penser que
les textes scéniques seraient plus proches de l’oral que les autres.

33 D enniston, Ixxv.
 YVESDUHOUX

Π se fait que les relevés effectués plus haut révèlent une propriété qui ne semble
pas encore été signalée jusqu’ici: les pièces de théâtre ont une proportion de parti­
cules significativement in fé r ie u r e à celles de tous les autres textes.
Ceci vaut d ’abord pour les pourcentages globaux des particules dans nos
oeuvres de référence:

te x te s n o n s c é n i q u e s :
Platon, M é n o n lA p o lo g ie d e S o c r a te : 16,83 % (3.080/1 δ.299);3435
Platon, P r o ta g o r a s : 16,86 % (2.924/17.342);
Xénophon, B a n q u e t: 18,31 % (1.726/9.424);35

te x te s s c é n i q u e s :
Aristophane, T h e s m o p h a r ie s : 12,94% (877/6.776);
Sophocle, O e d ip e à C o l o n e : 13,02 % (1.351/10.378).

La même constatation s’applique aux passages dialogués ou non de ces oeuvres. Le

tableau 6 permet de confronter les deux pièces de théâtre avec leurs correspondants
non scéniques. Le résultat de chaque comparaison se trouve dans la zone constituant
l’intersection des corpus comparés.

Platon, Platon, Platon, Platon, Xénophon, Xénophon,

Ménon Apologie Prot. dial. Prot. n o n Banquet Banquet
(dial.) (non dial.) dial. dial. non dial.
Aristophane, * * * * * 5*
Thesm. dial. Plat. Ar. Plat. > Ar. Plat. > Ar. Plat. > Ar. Xén. > Ar. Xén. > Ar.
Aristophane, * * * * * *
Thesm non Plat. > Ar. Plat. > Ar. Plat > Ar. Plat. > Ar. Xén.>Ar. Xén.> Ar.
dial.
Sophocle, * * Φ * *
O.C. dial. Plat. > S. Plat. > S. Plat. S. Plat. > S. Xén. >S. Xén. > S.
Sophocle, * * * « * *
O.C. n o n Plat. > S. Plat. > S. Plat. > S. P lat > S. Xén. > S. Xén. >S.
dial.

6. Comparaison de l’emploi des particules dans les parties dialoguées ~ non

dialoguées des oeuvres scéniques ~ non scéniques

34 Ces deux oeuvres sont regroupées de manière à fournir une répartition entre parties dialoguées ou
non comparable aux autres corpus.
35 Xénophon a un nombre significativement plus grand de particules que tous les autres corpus.
Nous y reviendrons plus loin (§ 8.2).
 GREC ÉCRIT ET GREC PARLÉ 33

À côté de ces différences quantitatives très nettes, la situation qualitative est la suivante
en matière d’éventail des particules utilisées dans les textes scéniques ou non:36
Aristophane et Sophocle ont respectivement 29 et 28 particules différentes, contre 30
dans le M é n o n et V A p o l o g i e , 28 dans le P r o t a g o r a s et 30 dans le B a n q u e t de
Xénophon. Sophocle ne diffère significativement d’aucun des trois corpus en prose.
En revanche, Aristophane a un choix de particules significativement plus étendu que
le M é n o n fV A p o lo g ie et que le P r o ta g o r a s ; il ne diffère toutefois pas du B a n q u e t.

5.2. Les particules d ans les textes versifiés scéniques ~ non scéniques

On doit se demander si la différence qui oppose les pièces de théâtre aux autres
oeuvres ne pourrait pas tenir non pas à leur caractère scénique, mais à leur forme
v e r s ifié e . En vue de vérifier cette hypothèse, j ’ai cherché à comparer la fréquence des
particules des pièces de théâtre avec des textes versifiés non scéniques d ’auteurs
attiques. En fait, le seul corpus disponible de taille suffisante me semble être celui des
poèmes de Solon (fin du VIIe s.—début VIe s.).37 On trouvera ci-dessous le tableau
des particules soloniennes (pour la procédure utilisée, voir § 3.3).

NOMBRE D ’OC­
CURRENCES
αλλά 9
0.00.
άοα
αΰ
νάρ 19
γε 1
γουν
δέ 108
6
δήπου
δήτα
η 3
η
καί 56
καίτοι
25
uévroi
urrôé 3
wiv

36 Voir note 29.

37 Édition utilisée par le TLG\ M. L. West, Oxford, 1972.
 YVES DUHOUX

urne 2
νυν
OUCÛC
ούδέ 15
οϋκουν
ούκοΰν
οΰν
οϋτε 9
τε 22
τοι 4
τοίνυν

7. Emploi des particules chez Solon

Le corpus solonien comporte 1.802 formes, parmi lesquelles figurent 282 particules
(15,64 % ) . La comparaison de ces données avec les textes scéniques examinés
antérieurement donne les résultats suivants:

Solon - Aristophane, T h e s m o p h o r i e s (dialogué; dialogué + non dialogué):

différences significatives (Solon > Aristophane).
Aristophane, T h e s m o p h o r ie s (non dialogué): différence non significative.

Solon - Sophocle, (dialogué; non dialogué; dialogué + non

O e d ip e à C o lo n e
dialogué): différences significatives (Solon > Sophocle).

En fait, les particules de Solon sont nettement plus fréquentes que celles des deux
pièces de théâtre versifiées—seule exception: les parties non dialoguées des T h e s ­
m o p h o r ie s d ’Aristophane, où la différence n’est pas significative.
On peut se demander si les deux siècles qui séparent Solon des deux autres
textes comparés ne pourraient pas avoir joué un rôle dans le contraste qui les oppose.
Ce facteur ne peut certainement pas être exclu en théorie, mais l’absence d’indice
précis en sa faveur invite à penser qu’il n’a pas exercé d ’influence démontrable. En
conclusion, ce n’est probablement pas la forme versifiée des textes qui influence la
fréquence globale des particules: c’est sans doute leur caractère scénique ou non.
On observera par ailleurs que la fréquence totale des particules soloniennes ne
diffère pas significativement de celle des trois textes platoniciens en prose.38 La com­
paraison avec le texte de Xénophon donne les résultats que voici: la proportion n’est

38 Le détail des comparaisons est le suivant:

Solon - Platon, Minore, Apologie; Ménon + Apologie: différences non significatives;
Solon ~ Platon, Protagoras (dialogué; non dialogué; dialogué + non dialogué):
différences non significatives.
 GREC ÉCRIT ET GREC PARLÉ 35

pas différente de celle des textes dialogués, mais est inférieure aux parties non dia-
loguées et à l’ensemble de l’oeuvre.

5.3. Fréquences individuelles des particules

Il est intéressant de déterminer si certaines particules ne jouent pas un rôle plus

important que d ’autres dans l ’usage propre aux textes scéniques ou non. Les deux
tableaux suivants permettent d’en avoir une idée précise. Ils donnent le résultat de la
comparaison de chacune des deux pièces de théâtre avec chacun des trois corpus non
scéniques (sans distinction entre parties dialoguées ou non).

PLATON, MÉNON PLATON, XÉNOPHON,

ET APOLOGIE PROTAGORAS BANQUET
άλλά = - =
άρα = - φns < s
αρα _ = NC
αΰ _ = =

Υάρ = = _
γε - Φ ns < s =

γοΰν NC NC =

δέ = Φ ns > s Φns > s

δη φ ns > s Φ ns > s Φ ns > s
δηπου = NC NC
δητα φ ns < s NC Φ ns < s
η φ ns > s φ ns > s Φ ns > s
η NC NC NC
καί φ ns > s Φ ns > s φ ns > s
καίτοι NC NC NC
UÉV Φ ns > s Φ ns > s Φ ns > s
μέντοι = = =

αηδέ NC NC NC
μήν NC NC Φ ns > s
μήτε - NC NC
νυν NC NC Φ ns < s
όμως NC NC NC
ούδέ = = =
ούκουν NC NC NC
ούκουν Φ ns > s Φ ns > s φns > s
οΰν Φ ns > s Φ ns > s =
ούτε Φ ns > s Φ ns > s =
τε » = =
 τοι NC * ns < s
τοίνυν = * ns < s

8. Comparaison de l’emploi des particules dans les T h e s m o p h o r ie s

d’Aristophane et les corpus non scéniques

a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o t s d a n s le s te x t e s s c é n i q u e ~ n o n
s c é n iq u e : voir § 5.1.
b) N o m b r e d e p a r t i c u l e s d i f f é r e n t e s d a n s le s te x te s s c é n i q u e ~ n o n s c é n i q u e : voir §
5.1.
c) P a r t i c u l e s n e d i f fé r a n t p a s d e fr é q u e n c e d a n s le s te x te s s c é n iq u e ~ n o n s c é n i q u e :
M é n o n t t A p o lo g ie : 13/30 (άλλα, άρα, άρα, αΰ, γάρ, γε, δέ, δήπου, μέντοι,
μήτε, ούδέ, τε, τοίνυν).
P r o ta g o r a s ·. 8/30 (αλλά, άρα, άρα, αύ, γάρ, μέντοι, ούδέ, τε).
B a n q u e t : 11/30 (αλλά, αΰ, γάρ, γοΰν, μέντοι, ούδέ, οΰν, οϋτε, τε, τοίνυν).
d) P a r tic u le s d i f fé r a n t d e f r é q u e n c e d a n s le s te x te s s c é n iq u e ~ n o n s c é n iq u e :
M é n o n et A p o lo g ie : 8/30 (δή, δήτα, ή, καί, μέν, ούκοΰν, οΰν, ούτε).
P r o t a g o r a s : 11/29 (γε, δέ, δή, ή, καί, μέν, ούκοΰν, οΰν, ούτε, τοι, τοίνυν).
B a n q u e t : 11/30 (άρα, δέ, δή, δήτα, ή, καί, μέν, μήν, νυν, ούκοΰν, τοι).

Ces dernières particules s’opposent comme suit:

Fréquence plus élevée à la scène:
M é n o n et A p o lo g ie : δήτα
P r o ta g o r a s : γε, τοι, τοίνυν.
B a n q u e t: άρα, δήτα, νυν, τοι.

Fréquence plus élevée dans les oeuvres non scéniques:

M é n o n et A p o lo g ie : δή, ή, καί, μέν, ούκοΰν, οΰν, ούτε.
P r o ta g o r a s : δέ, δή, ή, καί, μέν, ούκοΰν, οΰν, ούτε.
B a n q u e t: δέ, δή, ή, καί, μέν, μήν, ούκοΰν.

PLATON, PLATON, XÊNOPHON,

M É N O N ET PROTAGORAS BANQUET
APOLOGIE
αλλά = = =
άοα * ns > s _ =
άοα = = * ns < s
αΰ = φ ns > s =
γάρ * ns < s * ns < s =
γε * ns > s = =
νοΰν NC NC * ns > s
δέ * ns < s = * ns > s
 GREC ÉCRIT ET GREC PARLÉ 37

φ ns > s φ ns > s φ ns > s

δήπου φ ns > s NC NC
δήτα Φ ns < s Φ ns < s Φ ns < s
Φ ns > s Φ ns > s Φ ns > s
η φ ns < s φ ns < s =
καί φ ns > s Φ ns > s Φ ns > s
καίτοι = NC NC
ιιέν * ns > s Φ ns > s *ns > s
ιιέντοι φ ns > s /n s> s φ ns > s

μηδέ Φ ns < s Φ ns < s φ ns < s

- /n s< s φ ns > s

ιιήτε _ φ ns < s φ ns < s

νυν NC NC NC
ÖUCOC - NC =
ούδέ _ Φ ns < s Φ ns < s
ούκουν NC NC NC
ούκούν φ ns > s φ ns > s φ ns > s

οΰν Φ ns > s φ ns > s φ ns > s

ούτε = = Φ ns < s
τε Φ ns < s _ =
τοι φ ns < s φ ns < s Φ ns < s
τοίνυν φ ns > s NC Φ ns > s

9. Comparaison de l’emploi des particules dans I'O e d i p e à C o io n e

de Sophocle et les corpus non scéniques

a) E n s e m b l e d e s p a r t i c u l e s ~ le r e s t e d e s m o t s d a n s le s t e x t e s s c é n i q u e ~ n o n
s c é n i q u e s : voir § 5.1.
b) N o m b r e d e p a r t i c u l e s d i f f e r e n te s d a n s l e s te x te s s c é n iq u e ~ n o n s c é n i q u e s : voir §
5.1.
c) P a r t i c u l e s n e d if f é r a n t p a s d e f r e q u e n c e d a n s le s te x te s s c é n iq u e ~ n o n s c é n iq u e s :
M é n o n et A p o l o g i e : 9/30 (άλλα, άρα, αΰ, καίτοι, μήν, μήτε, όμως, ούδέ, ούτε).
P r o ta g o r a s ·. 7/30 (άλλα, άρα, άρα, γε, δέ, ούτε, τε).
B a n q u e t : 8/30 (άλλά, άρα, αύ, γάρ, γε, ή, όμως, τε).
d) P a r t i c u l e s d i ffé r a n t d e f r é q u e n c e d a n s l e s te x te s s c é n iq u e ~ n o n s c é n i q u e s :
M é n o n et A p o l o g i e : 18/30 (άρα, γάρ, γε, δέ, δή, δήπου, δήτα, ή, ή, καί, μέν,
μέντοι, μηδέ, ούκούν, ούν, τε, τοι, τοίνυν).
P r o t a g o r a s : 16/30 (αΰ, γάρ, δή, δήτα, ή, ή, καί, μέν, μέντοι, μηδέ, μήν, μήτε,
ούδέ, ούκοϋν, οΰν, τοι).
B a n q u e t : 18/30 (άρα, γοΰν, δέ, δή, δήτα, ή, καί, μέν, μέντοι, μηδέ, μήν, μήτε,
ούδέ, ούκούν, οΰν, ούτε, τοι, τοίνυν).
 YVES DUHOUX

Ces dernières particules s’opposent comme suit:

Fréquence plus élevée à la scène:
M é n o n et A p o lo g ie : γάρ, δέ, δήτα, η, μηδέ, τε, τον.
P r o t a g o r a s : γάρ, δήτα, ή, μηδέ, μήν, μήτε, ουδέ, τον.
B a n q u e t : δρα, δήτα, μηδέ, μήτε, ούδέ, ούτε, τοι.

Fréquence plus élevée dans les oeuvres non scéniques:

M é n o n et A p o lo g ie : άρα, γε, δή, δήπου, ή. καί, μέν, μέντοι, ούκοΰν,
ουν, τοίνυν.
P r o t a g o r a s : αύ, δή, ή, καί, μεν, μέντοι, ούκοΰν, ούν.
B a n q u e t : γοΰν, δέ, δή, ή, καί, μέν, μέντοι, μήν, ούκοΰν, οΰν, τοίνυν.

5.4. Conclusion

Le caractère théâtral ou non des textes a les conséquences suivantes sur l’emploi des
particules (les remarques ci-dessous ne mettent en jeu que les cas où le χ2 est calcu­
lable).

5.4.1. A l’exception de la comparaison des T h e s m o p h o r i e s avec un des corpus pla­

toniciens, il n ’a pas d’influence qualitative perceptible (§ 5.1).39

5.4.2. En revanche, les particules sont significativement m o i n s fréquentes au théâtre

qu’ailleurs—rappelons que l’opposition entre prose - vers ne joue pas de rôle per­
ceptible dans la fréquence globale des particules (§ 5.2).

5.4.3. Il existe une large minorité de particules à fréquence non différenciée (quatre
dixièmes des emplois où le χ2 est calculable:40 56/138), mais il y en a une majorité
sensible au caractère scénique ou non des textes (six dixièmes: 82/138).41 Les
préférences de ces dernières vont surtout dans le sens d’une plus grande fréquence
dans les textes non scéniques (52 ex.); l’inverse est moins courant (30 ex.). La nette
préférence globale des particules pour les oeuvres non théâtrales résulte donc de
l’interaction de trois tendances différentes.

5.4.4. Contrairement aux textes dialogués ou non (§ 4.6.3), les oeuvres scéniques ou
non scéniques s’opposent par un noyau de cinq particules différenciées dans tous
leurs emplois: δή, ή, καί, μέν et ούκοΰν.42 Ce petit ensemble joue un rôle important,
puisqu’il totalise pas moins de 46,7 % de toutes les particules (4.656/9.958). Il existe

39 Voir note 29.

40 Les observations qui suivent ne mettront en jeu que ce type d’emplois.
41 La différence entre majorité - minorité est significative.
42 II y en a trois autres qui ont des emplois différenciés dans cinq comparaisons sur six (δητα, οΰν
et τοι).
 GREC ÉCRIT ET GREC PARLÉ 39

une sixième particule à emploi constant: il s’agit (Γάλλά, qui est toujours indifféren­
ciée.43
Plusieurs autres particules ont des profils irréguliers: on en a 17/30 qui
témoignent de fréquences différenciée e t indifférenciée. Parmi celles-ci, il y en a
même six dont la différenciation s’opère en sens opposés: αρα, γε, δέ, μην, ούτε et
τοίνυν.
Six particules ont donc des comportements stables, mais le choix des fréquences
de plusieurs autres est largement arbitraire.

5.4.5. Toutes ces caractéristiques, constantes ou irrégulières, sont systématiquement

utilisées pour former dans chaque texte un groupe plus ou moins grand de particules
(de huit à dix-huit) dont la fréquence contribue à opposer les oeuvres théâtrales aux
autres. Leur liste est la suivante (entre parenthèses: le nombre d ’emplois différen­
ciés):

Particules plus fréquentes dans les oeuvres s c é n iq u e s : αρα (1 ex.), γάρ (2 ex.),
δητα (5 ex.), ή (2 ex.), μηδέ (3 ex.), μήτε (2 ex.), νυν (1 ex.), οΰδέ (2 ex.),
τ ε (1 ex.), t o i (5 ex.).

Particules plus fréquentes dans les oeuvres n o n s c é n i q u e s : αΰ (1 ex.), γούν (1

ex.), δή (6 ex.), δήπου (1 ex.), ή (6 ex.), καί (6 ex.), μέν (6 ex.), μέντοι (3
ex.), ούκοΰν (6 ex.), ούν (5 ex.).

Particules plus fréquentes dans les oeuvres s c é n i q u e s e t n o n s c é n i q u e s : αρα

(chaque fois 1 ex.), γε (chaque fois 1 ex.), δέ (3 ex. de >; 1 ex. de <), μην
(2 ex. de >; 1 ex. de <), ούτε (2 ex. de >; 1 ex. de <), τοίνυν (2 ex. de >; 1
ex. de <).

6. C onclusions générales

6.1. Le travail que voici ne pouvait bien entendu examiner que des particularités du
grec ancien écrit. J’espérais toutefois obtenir des indications indirectes sur ce que
devait réellement être l’emploi des particules dans le grec parlé en exploitant la carac­
téristique suivante: les textes grecs ne visent pas tous la même ressemblance avec
l’oral, mais certains tendent à lui être moins infidèles que d'autres. Il suffisait alors
de chercher s’il existait entre eux des différences significatives d’emploi des parti­
cules. Si tel était le cas, les usages propres aux corpus les moins éloignés de l’oral
refléteraient ceux de la langue parlée.
La recherche a porté sur deux caractéristiques des textes, leur nature dialoguée ou
non (§ 4) et scénique ou non (§ 5).

43 C e ci ne signifie évid em m en t pas q u 'e lle ne p o urrait pas différer dans d 'a u tre s échantillons.
Signalons que deux particules o n t des emplois indifférenciés cinq fois sur six: a u e t t e .
 YVES DUHOUX

6.2. Les contextes dialogués ou non n’ont pas d ’influence q u a l i t a t i v e perceptible,

puisque l’éventail des particules utilisées est le meme dans tous les cas (§ 4.6.1).44
En revanche, une partie des comparaisons entre oeuvres scéniques ou non révèle des
différences significatives: Aristophane a proportionnellement plus de particules dif­
férentes que les deux oeuvres de Platon (§ 5.1). Il n’y a cependant pas de différence
observable entre Aristophane et Xénophon, pas plus qu’entre Sophocle et les trois
corpus en prose.

6.3. La fréquence globale des particules n’est jamais plus grande dans les parties
dialoguées qu’ailleurs (§ 4.6.2). En revanche, elle est nettement moindre au théâtre
qu’ailleurs (§ 5.1).
Cette dernière caractéristique s’explique, à mon avis, par une propriété appar­
tenant exclusivement aux textes scéniques: le théâtre met en scène, c’est-à-dire qu’il
présente des acteurs qui se comportent d ’une certaine manière comme s’ils vivaient
réellement ce qui leur arrive. Il va de soi que ce ‘réalisme’ a ses limites: elles tiennent
aux coutumes du théâtre grec (masques; rôles féminins joués par des hommes; alter­
nance de paroles, chants et danses; etc.), ainsi qu’au caractère non authentiquement
oral du texte. Cette dernière caractéristique est spécialement frappante dans la
tragédie, dont Aristote reconnaissait qu’elle utilisait des tournures que l’on n’em­
ploierait jamais dans la conversation.45 Malgré toutes leurs conventions, la tragédie et
la comédie avaient cependant toutes deux un point commun essentiel: elles tendaient à
créer l’illusion de la réalité. Cette particularité n’a pas échappé aux Grecs: Platon
dans sa R é p u b l i q u e décrit explicitement tragédie et comédie comme faisant partie
d’une espèce ‘entièrement imitative’,46 ce qui implique, comme il le dit quelques
lignes plus haut, que l’auteur ‘essaie de nous faire croire dans toute la mesure du
possible que ce n’est pas [lui] qui parle, mais [son personnage].’47 Il semble que
l’impression de ‘vécu’ que donnait le théâtre grec n’ait pas été trop mauvaise, même
pour la tragédie, à en croire le récit par Hérodote des larmes qu’auraient versé les
spectateurs de la P r is e d e M i l e t de Phrynichos, prédécesseur d ’Eschyle.48 De cet
essai, même partiel, de rendre la réalité il découle que le théâtre grec mettait très
probablement en oeuvre davantage de ressources propres au langage oral que les
textes non scéniques.

44 Voir note 29.

45 Poétique 1458b- 1459a.
46 ... ή μέν διά μιμήσεως όλη έστιν, ώσπερ σύ λέγεις, τραγωδία τε καί κωμωδία: R ép u b liq u e
394b-c.
47 Ce passage porte sur l'épopée: πειράται ήμάς 0τι μάλιστα ποιήσαι μή "Ομηρον δοκείν είναι
τόν λέγοντα. αλλά τόν ιερέα (R épublique 393b). Sur le problème de la convention théâtrale dans
l’Antiquité, voir D. Bain, Actors a n d Audience, Oxford, 1977, 1-12. Dans le meme sens, voir le texte
de Longin cité par P. Ghiston-Bistagnc, Recherches sur les acteurs dans la G rèce a n tiq u e , Paris,
1976,3.
48 Hérodote 6.21 (ές δάκρυα ... έπεσε τό θέητρον). La V ita d ’Eschyle (§ 7) évoque, quant à elle,
des décès de petits enfants et des avortements qu’aurait suscité l’entrée en scène du chœur des
E um énides (τινές δέ φασιν έν xfj έπιδ είξει τών Ευμενιδών σποράδην είσαγαγόντα τόν χόρον
τοσοΰτον έκπληξαι τόν δήμον, ώς τά μέν νήπια έκψΰξαι, τά δέ έμβρυα έξαμβλωθήναι).
 GREC ÉCRIT ET GREC PARLÉ 4]

Nous ignorons, par définition, les caractéristiques propres au grec oral, mais
nous pouvons raisonnablement postuler qu’elles devaient comporter, au minimum,
l’intonation,49 plus une série de particularités syntaxiques analogues à celles que l’on
observe dans les dialogues oraux de bien des langues, sans compter des spécificités
lexicales ou phonétiques. Le tout s’enchâssait dans un contexte concrètement
appréhendable—gestes, présence des interlocuteurs ou auditeurs, environnement
matériel, situations politique, économique, psychologique, etc.—qui interagissait avec
la parole prononcée.
Tout ce qui vient d’être dit invite à admettre que la plus faible proportion de par­
ticules au théâtre résulte d’un plus grand usage de certaines caractéristiques du grec
parlé—pour un essai d’indentification de ces caractéristiques, voir § 6.5.
Sur l’idée, indéfendable, que les particules n’auraient existé que dans le langage
écrit, voir § 3.2. Rappelons que le caractère versifié ou non des oeuvres n’a pas
d ’effet démontrable sur la fréquence globale des particules (§ 5.2).50
Ce qui précède montre que, contrairement à ce que l’on croit souvent (§ 3.2), il
est sans doute inexact de penser que les particules étaient spécialement fréquentes
dans la langue parlée. C’est l’inverse qui est probablement vrai: les particules étaient
davantage utilisées à l’écrit. Une confirmation de ceci est fournie par l’usage de
Xénophon, qui emploie moins de particules dans les parties dialoguées du B a n q u e t
que dans les autres (§ 4.6.2,8.2; voir aussi § 7).
On peut désormais donner la réponse à une question soulevée par Denniston.
Observant que certaines associations de particules sont beaucoup plus rares (ou
même absentes) en vers qu’en prose, il constatait que ceci revenait à regrouper la
comédie et la tragédie et s’interrogeait avec perplexité sur les raisons de ce rap­
prochement qui lui paraissaient ‘not easy to find’.51 Après avoir pratiquement exclu
des facteurs métriques, il évoquait la possibilité de raisons diachroniques pour cer­
tains emplois. Je soupçonne que l’une des raisons de son embarras tient à son idée
qu’il aurait existé une opposition de fréquence des particules dans la comédie et la
tragédie, la comédie en ayant selon lui davantage que la tragédie. J ’ai montré ailleurs
que cette conception n’est pas défendable.52 Une autre raison tient probablement à
l’opposition trop tranchée que l’on a pris l’habitude d ’établir depuis Aristote entre
comédie et tragédie. En fait, toutes deux ont évidemment un point commun essentiel:
elles sont des textes scéniques et, de ce fait, tendent à ne pas trop éloigner leur emploi
des particules de celui du grec parlé.

6.4. Les auteurs disposent d ’une liberté impressionnante dans leur emploi des par­
ticules, mais ils l'exercent différemment dans les textes dialogués ou non et dans les
oeuvres scéniques ou non (§ 4.6.3, 5.4.4). Dans les premiers, aucune particule n’est
constamment différenciée, alors que dans les secondes, il n’y en a pas moins de cinq;

49 Sur l'intonation (ou accent de phrase) en grec ancien, voir Duhoux, § 7.

50 Pour le rôle du mène sur la fréquence de certaines particules, voir Duhoux, § 3.
51 Denniston, Ixxvii et n. 2.
52 Duhoux, § 5.
42 W E S DUHOUX

δή, ή, καν, μέν et οΰκοΰν. En revanche, άλλα est toujours indifférenciée dans tous
les corpus.
En fait, un nombre important de particules peuvent être différenciées e t indif­
férenciées: 15/30 (textes dialogués ou non); 17/30 (oeuvres scéniques ou non). Il en
existe aussi 15 qui peuvent avoir des différenciations allant en sens opposés. Ce
dernier phénomène se manifeste de deux manières. D’une part, des particules peuvent
être utilisées plus fréquemment à l a f o i s dans des textes dialogués (ou scéniques) e t
non dialogués (ou non scéniques): ή et μην (textes dialogués ou non); αρα, γε, δέ,
μέντοι, μην, ούτε et τοίνυν (textes scéniques ou non). Les particules peuvent aussi
être utilisées plus fréquemment dans des textes dialogués e t non scéniques (αΰ, δή,
οΰκουν et ουν), ou bien non dialogués e t scéniques (γάρ, ούδέ et τε).
Il faut conclure de tout ceci que bon nombre de choix de fréquence des particules
semblent arbitraires. La leçon à tirer dans ces cas est que, comme le fait prévoir la
théorie structuraliste, c’est l’opposition qui crée le sens.

6.5. Malgré leur emploi largement arbitraire, les variations de fréquence des particules
sont au service d’un projet cohérent: caractériser les contextes dialogués ou non, ainsi
que scéniques ou non. Ce résultat est obtenu par des groupes de 7 à 18 particules à
fréquences différenciées (§ 4.6.4, 5.4.5).
Les cinq seules particules à emploi différencié stable sont toujours plus
fréquentes dans les oeuvres n o n s c é n i q u e s . Du point de vue de leurs fonctions, ces
particules se répartissent comme suit:

(1) Particules pouvant fonctionner comme conjonctions de coordination: ή,

καί, μέν53 et ούκοΰν.

(2) Particule pouvant p o u v a n t fo n c tio n n e r c o m m e c o n j o n c tio n de coordina­

tion et comme modalisateur (sens intensif): δη.

Ces cinq particules ont une préférence pour les textes qui cherchent le moins à
ressembler à la langue parlée. Ceci fait conclure qu’elles tendaient probablement à
être moins utilisées à l’oral qu’à l’écrit. La préférence de καί pour les oeuvres non
scéniques est à mettre en parallèle avec son attirance marquée pour les textes non dia­
logués (4 ex. sur 5):54 les deux vont manifestement dans le même sens. Le
phénomène est d’autant plus intéressant que καί est de loin la particule la plus
fréquente de toutes;55 elle restera fort employée dans le N o u v e a u T e s t a m e n t et dans
le grec écrit postclassique.56

53 Μέν peut aussi fonctionner comme modalisateur (sens intensif), mais elle sert le plus souvent de
conjonction de coordination.
54 Observer que tel est aussi le cas de μέν (3 ex. sur 5).
55 Dans les corpus étudiés ici, καί totalise 3.027 occurrences, soit un peu plus de 30 % des 9.958
particules prises en compte; la particule suivante est δέ, attestée seulement 1.470 fois. Voir aussi § 2.
56 Blotnqvist, 133-141; Schwyzer II, 634.
 GREC ÉCRIT ET GREC PARLÉ 43

Π faut se demander maintenant quelle est la différence que les cinq particules ci-
dessus mettaient en oeuvre entre oral et écrit. Π ne s’agit bien sûr pas de l’intonation,
comme on aurait peut-être pu le croire a p r i o r i , mais de la syntaxe, puisque la plupart
de ces particules sont des conjonctions de coordination. Ceci suggère que le grec
parlé avait tendance à d a v a n t a g e d ’a s y n d è t e que le grec écrit.57 Cette conclusion
mériterait des développements que je ne puis donner ici, aussi je me borne à signaler
un seul fait parallèle. Il concerne l’emploi, dans nos cinq corpus de référence, de six
conjonctions de subordination choisies arbitrairement (il s’agit d’èreei, επειδή, ίνα,
δπως, ότι et ώστε).58 Leur comparaison donne le résultat suivant: dans les deux
pièces de théâtre, leur fréquence est significativement in fé r ie u r e à celle des trois cor­
pus non scéniques.59 Ici aussi, par conséquent, il existe de nets signes de préférence
des textes les plus proches de l’oral pour l’asyndète.
Il est intéressant de comparer ces données avec les conclusions de S. Trenkner
dans son enquête sur le ‘style κ α ί’. Rappelons que cette forme d ’expression se
caractérise surtout par une forte fréquence de propositions successives introduites par
καί.60 L ’étude porte sur l’emploi de καί dans le récit dit ‘simple’, jugé ‘fidèle à la
tradition syntactique orale’—par opposition au récit ‘artistique ... qui se soumet aux
exigences esthétiques de l’époque’.61 Trenkner n’a malheureusement comparé ni la
fréquence de καί dans les récits ‘simples’ ~ ‘artistiques’,62 ni l’emploi de la coordi­
nation ~ subordination dans les récits ‘simples’. Ceci limite la portée de ses conclu­
sions, qui sont les suivantes pour le point qui nous intéresse: le ‘style κ α ί’ est
‘typique et très usuel du récit oral’;63 il ‘indique, à l’état d’ébauche, le mécanisme de
la parataxe parlée de l’époque’.64 Il était ‘en usage dans toutes les couches sociales,
depuis le milieu de charcutier jusqu’à Platon’.65 Pour Trenkner, donc, le ‘style κ αί’
est une marque du grec parlé. La conclusion de mon examen est que l’emploi de καί
est spécialement fréquent dans les textes l e s m o i n s p r o c h e s d e l ’o r a l —cette
différence n ’étant pas négligeable: le P r o t a g o r a s de Platon ou le B a n q u e t de
Xénophon ont d e u x f o i s p l u s de καί que 1'O e d ip e à C o lo n e . . .
Il y a bien entendu contradiction entre les deux études, mais je pense qu’elle
n’est qu’apparente. En effet, le travail de Trenkner ne porte que sur une partie très
réduite des occurrences de καί. Les seuls emplois communs à nos deux études sont
les suivants: dans les T h e s m o p h o r i e s , Trenkner en a 6 (six)—2,6 % de mes 231

57 Déjà en ce sens, Schwyzcr II, 634.

58 Les occurrences de ces six conjonctions représentent près du tiers des subordonnants recensés par
Cauquil-Guillaumin.
59 II n’y a de différence significative ni entre les deux œuvres scéniques ni entre les trois corpus
non scéniques.
^T ren k n er,8-11.
61 Trenkner, 2-3.
62 Pour d’autres emplois du ‘style κ α ί’, voir Trenkner, 10-12.
63 Trenkner, 5.
64 Trenkner, 59.
65 Trenkner, 78.
44 YVES DUHOUX

exemples; dans le P r o t a g o r a s , elle en a 99—9,6 % de mes 1.026 exemples.66 Le

diagnostic de Trenkner ne vaut donc p a s pour l’ensemble des emplois de καί, mais
seulement pour l’un de ses sous-groupes—elle-même signale d’ailleurs explicite­
ment que le ‘style καί’ ‘rivalisait avec d ’autres formes populaires, λέξις είρομένη et
style asyndétique’.67 On peut donc sans difficulté admettre simultanément que: 1) les
textes plus proches de l’oral ont une tendance générale à utiliser καί m o i n s s o u v e n t
que les oeuvres qui en sont plus éloignées, ce qui traduit une préférence du grec parlé
pour l’asyndète; 2) dans la petite fraction de ses emplois constituée par le ‘style καί’,
καί se trouve tr è s s o u v e n t dans le style oral.

6.6. Il faut rappeler que les conclusions ci-dessus ne valent que pour les corpus
examinés. Il serait intéressant d ’étendre l’enquête pour savoir si elles se vérifient
ailleurs (mais voir § 7). Je rêve du jour où nous disposerons de dépouillements
exhaustifs et rigoureux donnant la fréquence des particules de toutes les oeuvres aux
diverses époques de l’hellénisme. La réalisation de cet objectif est aujourd’hui à notre
portée. Elle permettrait de dresser enfin un tableau complet de l’emploi des particules
et d ’en retracer l’évolution. J’espère que les premiers résultats de la méthode présen­
tée ici inciteront d’autres chercheurs à aller dans cette direction.

7. P articules e t langue parlée: une vérification

L’étude que voici était déjà complètement rédigée lorsque j ’ai pu prendre connais­
sance de l’article de Dickey. Examinant les emplois du vocatif chez Aristophane et
Ménandre, elle conclut que l’usage de Ménandre serait largement conforme à la con­
versation attique, alors que celui d ’Aristophane serait plus sophistiqué. Ce jugement
me paraît fondé. Du coup, nous avons la possibilité de mettre à l’épreuve la conclu­
sion formulée à l’instant (§ 6.5), selon laquelle le grec parlé mettrait en oeuvre moins
de particules que le grec écrit. En effet, si Dickey a raison, on devrait s’attendre à ce
qu’Aristophane ait davantage de particules que Ménandre, puisque le premier est plus
éloigné de l’oral, alors que le second en est plus proche.
J’ai donc relevé les particules utilisées dans le Ô y s c o l o s de Ménandre68 (daté de
317/316 avant notre ère). Le dépouillement a été effectué selon la même procédure
que les autres textes (§ 3.3), à ceci près que les restitutions de l’édition ont été inté­
gralement conservées. Voici le résultat de cet examen:

66 Le corpus classique de Trenkner est d'ailleurs trois fois plus petit que le mien (899 ex. de καί,
contre 3.027). Elle y a inclus des extraits des Caractères de Théophraste, qui est postérieur d’au
moins deux générations à mes auteurs—or, Théophraste représente 39,1 % de ses emplois de καί,
avec une fréquence significativement supérieure à celle de ses autres auteurs (à l’exception du
Contre Olympiodore du Pscudo-Démosthène).
67 Trenkner, 78.
68 Édition utilisée par le TLG: F. H. Sandbach, Oxford, 1972.
 GREC ÉCRIT ET GREC PARLÉ 45

NOMBRE D ’OC­
CURRENCES
άλλα 62
αρα 2
αοα 6
αΰ 1
Ύάρ 84
37
νουν 3
δέ 141
20
δήπου
δ τ ίτ α 1
ή 12
η 2
καί 113
κ α ίτ ο ι
uév 24
ιχ έ ν τ ο ι 1
ι ιπ δ έ 5
2
αήτε 2
νυν
OUCOC 5
ουδέ 25
ούκουν
ούκοΰν 1
οΰν 8
ο ϋ τε 14
τε 47
τοι
το ίν υ ν 2

10. Emploi des particules dans le

de Ménandre
D y s c o lo s

Le D y s c o l o s comporte 6.217 formes, parmi lesquelles figurent 620 particules (9,97

% ).Cette fréquence est significativement plus basse que les 12,94 % des T h é s -
m o p h o r ie s d ’Aristophane (et que les 13,02 % de 1’O e d ip e à C o lo n e de Sophocle).
46 YVES DUHOUX

Comme cette différence est conforme à ce que faisait attendre la théorie, il se vérifie
que, à l’oral, le grec utilisait réellement moins de particules qu’à récrit.
Par ailleurs, j ’ai suggéré plus haut que la plus faible proportion de particules
dans la langue parlée refléterait une asyndètepLus grande qu’à l ’écrit. Cette idée avait
été étayée par l ’examen de l’emploi de six conjonctions de subordination choisies
arbitrairement. La comparaison de ces conjonctions dans le Dyscolos et les Thes-
mophories montre qu’Aristophane en utilise un nombre significativement plus élevé
que Ménandre. Ici encore, donc, le nouvel examen valide l’hypothèse.
On a parfois supposé que le processus de déclin des particules à l ’époque hel­
lénistique serait déjà visible chez Aristote et Ménandre.69 Ceci n’est en tout cas pas
soutenable pour Aristote: j ’ai montré ailleurs que sa Physique a une fréquence de
particules plus élevée que celle des oeuvres platoniciennes de mon corpus.70
En ce qui concerne Ménandre, il y a, de fait, faible proportion de particules.
Toutefois, comme ce phénomène va de pair avec un emploi nettement limité des con­
jonctions de subordination, on peut raisonnablement conclure que la basse fréquence
des particules s’explique par une prédilection pour l ’asyndète, typique du grec oral.

8. Annexe: Particules et auteurs

8.1. Généralités

Les auteurs s’opposent de deux façons les uns aux autres par leur fréquence de par­
ticules:

a) Les fréquences individuelles des particules peuvent varier d ’auteur à auteur: ainsi,
lorsque le χ2 est calculable, elles s’opposent 8 fois sur 16 dans les parties dialoguées
ou non de Platon, MénonlApologie ~ le Banquet de Xénophon; 4 fois sur 11 dans
celles de Xénophon, .S a lu e r “ Aristophane, Thesmophories; etc. Je n’ai pas effectué
toutes les comparaisons possibles, mais les deux taux de divergence ci-dessus ne sont
pas significativement différents de celui observé entre les deux corpus platoniciens
lorsque le χ2 est calculable (§ 4.6.3).

b) Les auteurs peuvent aussi s’opposerpar la fréquence de l'ensemble de leurs par­

ticules.

La comparaison révèle des différences significatives entre les auteurs suivants (pour
les données chiffrées, voir § 5.1,7): Platon a moins de particules que Xénophon, mais
plus qu’Aristophane, Sophocle et Ménandre; Xénophon et Ménandre en ont respec­
tivement plus et moins que tous les autres. Les deux corpus platoniciens ne se dif­
férencient pas, ce qui ne demande pas de commentaire. Le nombre relativement faible

69 R éférences dans Blom qvist, 132.

70 D uhoux, § 4.
 GREC ÉCRIT ET GREC PARLÉ 47

de particules des pièces de théâtre va dans le sens de la spécificité scénique signalée à

l’instant (§ 5, 6.3). La différence entre Ménandre, d’une part, et Aristophane et
Sophocle, de l ’autre, est due à ce que Ménandre est plus proche de la langue parlée
qu’Aristophane (§ 7) et sans aucun doute aussi que Sophocle. Sur le Banquet de
Xénophon, voir § 8.2.

8.2. Le B a n q u e t de Xénophon

Comme les autres textes étudiés, le Banquet de Xénophon oppose les textes dialogués
ou non par les fréquences différenciées d’un petit noyau de particules. Il diffère
toutefois des autres corpus par deux usages spécifiques.
D ’abord, il est le seul à utiliser davantage de particules dans ses parties non dia­
logues que dans les dialogues (alors que leur fréquence est globalement indifféren­
ciée dans les autres corpus). Ensuite, ses particules sont plus fréquentes que celles
des autres corpus étudiés, aussi bien dans les parties dialoguées que dans les
autres.7172Comment ces deux pratiques s’inscrivent-elles dans les usages des autres
auteurs examinés?
Pour la plus grande fréquence de particules dans les parties non dialoguées,
Xénophon innove: il ne se contente pas de jouer sur les fréquences d’un petit nombre
de particules, mais il les augmente substantiellement plus que les autres auteurs. Je
pense que cet usage est à rapprocher de l’opposition détectée entre textes scéniques ~
non scéniques (§ 5, 6.3, 7). Tout comme les particules sont plus nombreuses dans les
textes non scéniques qu’au théâtre, de même, Xénophon en utilise davantage dans ses
parties non dialoguées que dans ses dialogues. Il introduit donc à l ’intérieur d’une
oeuvre non scénique le procédé de caractérisation qui opposait les pièces de théâtre
aux autres textes. Et de façon assez compréhensible, il traite les dialogues comme les
pièces de théâtre—il leur confère donc un caractère plus oral. Si cette interprétation se
révélait correcte, elle jetterait une lumière nouvelle sur Xénophon, souvent jugé piètre
styliste—on a encore récemment spéculé sur son ‘artistic incapacity’P2
L ’élévation générale de fréquence de toutes les particules du Banquet constitue,
quant à elle, une utilisation intensive d ’une ressource que d’autres auteurs utilisent
avec davantage de réserve. Sur les conséquences à tirer de cet usage du point de vue
stylistique, voir Duhoux, § 4.

9. Principaux sigles utilisés

d : dialogué
NC : différence non évaluable par le test du χ2
nd : non dialogué
ns : non scénique
s : scénique

71 Une fréquence aussi élevée que lâ sienne se trouve dans la Physique d’Aristote, voir Duhoux, § 4.
72 V. Bers, G reek P oetic S yn ta x in the C lassical A ge, New Haven -Londres, 1984,13.
 YVES DUHOUX

= : difference non significative d ’après le test du χ2

Φ : différence significative d’après le test du χ2
> : significativement plus grand que (d’après le test du χ2)
< : significativement plus petit que (d’après le test du χ2)
~ : en regard de

Abrévìations bibliographiques

Blomqvist, J.
1969 Greek Particles in Hellenistic Prose. Lund
Cauquil, G. - J.-Y. Guillaumin
1985 Vocabulaire de base du grec. Besanpon
Denniston, J.D.
19542 The Greek Particles. Oxford
Dickey, E.
1995 ‘Forms of Address and Conversational Language in Aristophanes and
Menander’. Mnemosyne 48,257-271
Duhonx, Y.
1997 ‘Quelques idées rcques, et néanmoins fausses, sur les particules grecques’.
L’Antiquité Classique
Fraenkel, JJ.
1947 ‘A Question in Connection with Greek Particles’. Mnemosyne 13,183-201
Hellwig, A.
1974 ‘Zur Funktion und Bedeutung der griechischen Partikeln’. Gioita 52,145-
171
Labéy,D.
1950 Manuel des particules grecques. Paris
Meillet, A.
19657 Apergu d'une histoire de la langue grecque. Paris
Ruijgh, C.J.
1971 Autour de τε épique'. Études sur la syntaxe grecque. Amsterdam
Schwyzer, E.
1939-71 Griechische Grammatik. Munich
Sicking, C.M.J.
1986 ‘Griekse partikels: definitie en classificatie’. Lampas 19,125-141
Sicking, C.M.J. - J.M. Van Ophuijsen
1993 Two Studies in Attic Particle Usage. Lysias and Plato. Leydeetc.
TLG: Thesaurus Linguae Graecae (Université de Califomic à Irvine).
Trenkner, S.
1960 Le style καί dans le récit attique oral. Assen, 1960 [réimpression de l ’édi-
tion de 1948]
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS
The Case o f Arcadian

ANNA MORPURGO DAYffiS

Somerville College, Oxford

X. Three preliminary points

Three points are relevant to most discussions of Greek particles. The first concerns
the widespread belief that Greek is unbelievably rich in particles and combinations of
particles. Admittedly it is normally assumed that this applied to Homeric and Classi­
cal Greek and it is accepted that the position changes when we reach the New Testa­
ment Similarly it is taken for granted that different literary genres made different use
of particles. Denniston ([1934] 1954) carefully distinguishes between poetry and
prose. Yet we do not normally ask ourselves how rich in particles the dialects are, or,
less optimistically, what we know about the use of panicles in the dialects other than
Attic and Ionic. Denniston’s references to dialects concern literary dialects and
mainly Ionic and Attic.

The second point concerns the development of particles and of the use of particles in
Greek. The traditional but valuable S y n ta x e g r e c q u e by Humbert (1954: 370), just to
take an example, has a clear statement on the subject. The author justifies his decision
to discuss the Attic use of particles in the Fifth and Fourth Centuries B.C. pointing
out that it is far more developed and refined than anywhere else. ‘Chez Homère, le
clavier des particules est fort réduit, non seulement parce que les particules sont elles-
mêmes peu nombreuses, mais surtout parce qu’elles ne sont que grossièrement dif­
férenciées et ne se sont pas augmentées des nombreuses “combinaisons” que nous
atteste l’attique. Le développement des particules, l ’affinement de ces instruments
d’expression de plus en plus précis, est parallèle au développement, si rapide et si
brillant, de l’esprit grec entre le VIe siècle et la fin du Ve.’ One may disagree with the
general conclusion but the implication is that there is development and variety in the
use of particles. Is this development typical of dialects too? What evidence do wc
have?

A third and more general point has to do with the grammatical status of particles.
There is an unstated agreement that they belong to the systemic part of grammar; they
are a closed rather than an open class. Hence they are less likely to be borrowed and
 ANNA MORPURGO DAVIES

when due to innovations more likely to come into existence through the internal
development of the language. In other words, we like to believe that particles may
arise through complex processes of grammaticalization just as complementizers,
auxiliary verbs or other morphological categories. This belief is widely shared. Even
at the very beginning of the nineteenth century in the so-called prescientific linguis­
tics John Jamieson, the author of Hermes Scythicus or, The Radical Affinities o f the
Greek and Latin Languages to the Gothic (Edinburgh 1814) argued that particles
were the ideal ground on which to test the possible kinship of languages since they
were likely to be inherited and not borrowed: ‘The particles, or “winged words”, as
they have been nominated, are preferred in the proof of this affinity for several rea­
sons. These are generally of the highest antiquity .... They are also more permanent
than most other terms .... They are also least likely to be introduced into another lan­
guage ...’ (op. cit., ii: 2). Indeed, the view that borrowing is much more likely for
open classes like lexical items than for closed classes has long been accepted
(Haugen 1950). Should we then reckon that particles, which in all likelihood form a
closed class, are not normally borrowed? And docs this apply to particles as such (i.e.
to the possibility of particles as loanwords) or to the way in which they are used (i.e
to the possibility of semantic loans)? The answer is important for the Greek dialects.
If particles and particle use tend to be developed through internal mechanisms we
might expect considerable divergences between the dialects. More exactly, we might
expect such divergence, if a) the assumption just made about particles as belonging to
a closed class is correct, b) borrowing processes between dialects in general or
between the Greek dialects in particular are comparable to those between languages.2

2. Particles in inscriptions

We may now turn from the general to the specific. First (§ 2.1) I shall make some
general observations about the epigraphical use of particles; then I shall briefly dis­
cuss the data we have about particles in Arcadian and Cypriot (§ 2.2); this survey of
the material will finally lead me to concentrate on Arcadian and on the use of connec­
tives in Arcadian inscriptions (§ 3). Mycenaean will be used to provide some sort of
background to our history. It is with some trepidation that I offer this paper to Kees
Ruijgh, who has been both a friend and a model for thirty years and more, and who
has never ceased to impress and astonish me with his complete mastery of all forms
of Greek, early and late, literary and epigraphical.

2.1. Some general observations

That of particle is a vague concept and I cannot avoid any longer a statement about
what is meant by particle in this paper. I shall not try to provide a definition but shall
exploit a previously formulated list. For Attic Humbert (1954: 374ff.) mentions
αλλά, Spa, apa, ά-cap, αύτάρ, αΰ, αυτε, αΰτις, γάρ, γε, δαί, Sé, δή, ή, ή, καί,
καίτοι, μάν/μήν, μέν, μέντοι, οΰν, περ, που, τε, τοι, τοιγάρ, τοιγάρτοι,
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 51

τοιγαροΰν, τοίνυν. I shall use this list as a starting point, accepting Humbert’s
exclusions, i.e. I shall not consider, for instance, the so called potential particles like
αν, κε(ν), nor shall I deal with conjunctions like εί, έπεί, ώς, or prepositions like
επί, παρά, etc. However, I shall add to Humbert’s list the Homeric particles Ιδέ, νυ,
βην, and keep in mind the existence of dialect-specific particles like the Cypriot παν
or the Thessalian μα, which corresponds in usage to δε, What do we know about
theseparticles outside literature?
The answer is that we know very little; outside literature most particles are not
attested or are badly attested. Indeed the extent of this non-attestation must surprise
and the point requires further illustration.

I start with a body of material which ought in theory to be rich in particles—that of

Greek verse inscriptions. P.A. Hansen’s edition of the C a r m i n a e p i g r a p h i c a g r a e c a
Ι-Π (1983-89), even if not completely up-to-date, offers sufficient evidence for the
verse inscriptions written before 300 B.C. and I base my observations on it.
Particularly in the first volume (inscriptions dated before 400 B.C.) we find texts
heavily influenced by the epic language and we might expect that on the one hand the
desire to imitate epic poetry and on the other the need to fill slots in the metre may
have led to large use of particles which offer convenient monosyllabic or disyllabic
elements. A quick reading leads to different conclusions. Not even half of our list is
represented in the verse inscriptions of the period before 400. If we consider
Hansen’s second volume which contains for the most part inscriptions of the fourth
century the evidence increases but we still have considerable gaps. In any case the
main point is the rarity of these particles in the early period at least. In ca. 500 verse
inscriptions written before 400 only καί, δέ and τε or τ ε ... καί can be said to occur
reasonably frequently; for the rest άλλα, γάρ, μεν occur more than 8 times each,
while eight other particles (αΰθε, αύτάρ, είτε, εϋτε, η, ήδέ, περ, πω) occur at best
four times each but often once or twice each only.1 Obviously the absence of some
particles may be due to chance. Thus Hansen has no examples o f μάν, but an
instance of the particle occurs in the elegiac distichs of the Sixth Century Polyan­
drion of Ambracia recently published: ΚαΕ μάν Άραθθίονα KaE Εΰξενον ϊστε,
πολΐταΕ, / Ιιός μετά τδνδ’ άνδρόν Κάρ έκιχεν θανάτο.212

1 Hansen lists 465 inscriptions in CEG I and another 32 early inscriptions in CEG II (cf. p. 299).
The figures for the least frequent particles are given below distinguishing occurrences in Attic (A)
and in non-Atlic (nA) inscriptions: αλλά (A x 2; nA x 7), αΰθε (nA x 1), αύτάρ (A x 1; nA x 3),
γάρ (A x 6; nA x 4), ε ίτε (A x 1), ε ίτ ε (A x 1), fj (A x 1; nA x 2), ήδέ (A x 3; nA x 1), μέν (A
x 11; nA x 2 or 3), περ (nA x 2), πω (nA x 1). I have not counted the particles which are entirely
restored or doubtful.
2 For the text (five distichs which belong to the sixth century B.C.) see Andreou (1986), Bousquet
(1992), Cassio (1994); presumably in the name Ά ραθθίσ να the first vowel is long and the v is
consonantal, so that we have three long syllables followed by a short one. I owe to Albio Cassio
the observation that here the use of μαν is entirely parallel to that of Homer, where it regularly
occurs before vowel. On the other hand, as he points out, Homer has ού μάν before vowel as well as
ή μάν before vowel, but never κα ί μάν, while καί μέν occurs frequently before consonant and
rarely before vowel.
52 ANNA MORPURGO DAVIES

Even if new discoveries bring new evidence the rarity of particles is not in question;
the lesson to be learned at this stage is that even in poetry we need not expect the
same richness in particles or combinations of particles that we find in Homer or in
Attic literary prose. If that is so in verse inscriptions, we should not be surprised
when we turn to prose inscriptions in a local non-literary dialect. I have chosen Arca­
dian as a test case on various grounds. First the available data are of a manageable
proportion; secondly the dialect inscriptions have been recently edited by Dubois
(1986) and even more recently most of them have been reconsidered by Thür and
Taeuber (1994), so that one can use these texts with a few basic additions as an ade­
quate representation of what we have for the dialect; thirdly we may legitimately com­
pare Arcadian with Cypriot on the one hand and with Mycenaean on the other to
extend our historical perspective.

2.2. Arcadia and Cyprus

Arcadia as all other Greek regions has yielded a number of dialect inscriptions and a
greater number of inscriptions in koine; there are also a few instances of texts in
koina. The dialect inscriptions which concern us start in the sixth century B.C. and
continue down to the third century B.C. when the dialect is replaced by the koine.
What follows concerns the dialect inscriptions only.
As we might have expected the Arcadian inscriptions are not rich in particles: a
complete list (which again excludes modal particles and conjunctions) comprises:
αλλά (2 or 3 times), δέ (μηδέ, ουδέ), ή, κάς/καί, μέν, τ ε ( ζ \ είτε/ ε'ίΗε, μήτε,
ούτε), of which only δέ, ή, κάς/καί have any frequency. Γη addition oaths are intro­
duced by νεί, equivalent to vai found elsewhere in Greek, and there is a very doubt­
ful instance of άτάρ (see below).3
This paucity of particles is not, as I said, unexpected. It is striking, however, how
Arcadian differs from Cypriot not in the number but in the types of particles. In
Cypriot we do not seem to find either τε or δέ as such, though two occurrences of
m e - t e could be interpreted as including either δέ or τε.4 The basic connective is κάς
(not καί as in the majority of the Arcadian inscriptions), which is clearly used where
elsewhere we might expect δέ. In addition Cypriot offers evidence for three appar­
ently archaic particles: παι, ίδέ and αΰτάρ; the first is Cypriot only, the second and
third are also found in Homer and in the poetry influenced by the epic language but

3 For the sake of completeness, though it does not feature in Humbert's list, I add here the two
examples of ποτέ in two Jate Hellenistic defixiones sometimes attributed to Arcadia (Dubois 1986:
ii, 319ff.). In fact, as pointed out by Dubois, there is no real evidence thatlhe texts are Arcadian. On
the other hand I have not considered the particles δαν, καν, ναν identified by Dubois (1986: i,
227ff.), both because, as indicated above, I am not concerned with potential particles, and because I
follow Hodot (1990) in a different interpretation of the texts. Cf. also Dunkel (1990).
4 me-te ... me-te occur twice, in fourth century Paphos: ICS2 8, Masson-Mitford 1986, no. 237; for
Masson the correct interpretation is μήτε, but for Risch (1988: 73 note 22) it is μηδέ. In addition
an example o f ie, which may (but need not) be interpreted as τε or δέ occurs in ICS2, 406, and for
metrical reasons Neumann (1975: 154 = 1994: 533) introduces a δέ in the final line of ICS2 264.
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 53

not elsewhere.5 Other absences arc not significant given the limits of our evidence,
but even in our scrappy texts we would have expected to find a reasonable number of
occurrences of τε and δέ, if they had been in real use. The presence of ίδ έ and
αύτάρ is significant, since, as C.J. Ruijgh pointed out long ago (1967: 29-57), they
must belong to an earlier ‘Achaean1stratum. If so, it seems likely that Arcadian had
them too at some stage but presumably lost them or used them at a stylistic level or a
register which differs from that of our texts. Notice, however, that for Arcadian and
Cypriot we assume an earlier unity at a post-Mycenaean period (Morpurgo Davies
1992). This means that the differences which we have observed in the particle lists
must be due to developments which follow the period of unity. In other words as far
as particles are concerned the patterns of development are relatively fast.

3. A closer look at Arcadian

3.1. άτάρ, άλλα

We may now look more closely at the Arcadian list. The supposed άτάρ occurs only
once—and only if we accept a specific interpretation and reading of the text in ques­
tion (IG V 2 343, 50; cf. Dubois 1986: ii, 157f.). The mostrecent editors (Thür and
Taeuber 1994:133 atl. 47) do not and it may be wiser to suspend judgement.
Of the other particles αλλά is very rare: one or two examples from the fourth
century (in texts where outside influence cannot be excluded)6 and one later example
in a defixio which need not be Arcadian (Dubois 1986: ii, 320).

Two particles only, κάς or κ α ί and δέ, are reasonably frequent; the remaining ones,
μέν and τε are not. All of these require further discussion.

I hardly need to rehearse the history of τε; in its inherited labiovelar form (q e ) it is
omnipresent in Mycenaean but in alphabetic Greek it is mostly replaced by καί (or
κάς in Cypriot and parts of Arcadian). This is interestingly enough a panhellenic
phenomenon and the history of τε everywhere, including Arcadian, cannot be wholly
dissociated from that of κάς/καί, largely because its scarcity of occurrence is deter-

5 For the references cf. the Cypriot lexica by Hintze (1993) and Egetmeyer (1992) s.vv. pa-i, i-te,
au-ta-ra/a-u-ta-ra. It is likely, but obviously not certain, that the Idalion inscription (ICS2 217)
has a preform o f the Attic ουν and conceivably an instance o f ye; the ν υ particle seems to be
attested as well. Also η is found more than once, while a late verse inscription (ICS2 264) contains
a sequence o-wo-ka-re-ti where different exegeses have recognized the particle γάρ (Masson, ad ioc.)
or the particles αρ έτι (Neumann, loc. cit. in note 4).
6 A clear-cut example of άλλό is found in Schwyzcr DGE 6 27,27 (Dubois 1986: ii, 61ff.; Tegea’s
decree about the return of the exiles after Alexander the Great); the other example (IG V 2 6, 34,
Dubois 1986: ii, 40), in a set of rules about working contracts also from Tegea, belongs to the
notoriously difficult sequence άλλ' ή, where it is not even certain that άλλα (rather than άλλο) is
included (... μη οι έστω Ινδικόν μηδέποθι άλλ' η ίν Τεγέαι).
54 ANNA MORPURGO DAVIES

mined by the success of κάς/καί. In fact in Arcadian the best evidence for τε is in
some forms of compounds: είτε, μήτε, ούτε are all attested, even if very rarely, in
Arcadian inscriptions. Yet these are fossilized forms; they are not sufficient to indi­
cate that copulative τε as such is part of current usage.
On its own τε (or ζ’) occurs three times in dialect inscriptions, always in ... τε ...
καί constructions:

(1) Dubois (1986: ii, 196, possibly from Pheneos, ca. 500):

[κακο]ς ζ’ έξόλοντυ κά οζις τότε δαμιο^οργέ [άφάε]σται ... ‘let him perish
horribly and let whoever is then d a m i o r g o s pay ....’

(2) IG V 2 262,19 (Mantinea, fifth century)

α τε θεός κάς οί δικασσταί ‘the goddess and the judges1

(3) IG V 2 343,49 (Orchomenos, fourth century)

τάς τε iv Β[ΰ]αίμονι καί τάς i[v Έρχομινο]ΐ ‘those in Euaimon and those in
Orchomenos’

The particle is also found in an obscure and probably mistaken construction in the
Cos A s y l i e u r k u n d e n where it is not clear whether we deal with an example o f ... τε ...
καί ... καί joining nouns or (more likely) with an example of τε ... τε joining
clauses:

(4) Dubois (1986: ii, 230, dialect of Thelphousa, 242 B.C.)

... π α ρ εκ ά λ εα [ν ] τας τε θυσίας κοννανήν ... κα ί του άγώνος καί τας

έκεχηρίας τό τ[ε] ιερόν άσυλον η ν α ι... ‘invited them to share in the sacrifice
... and in the agon and the truce, and to let the sanctuary be inviolable ...’

Given the absence of τε in Cyprus and the rarity of the Arcadian attestations we may
feel tempted to assume that τε survived in Arcadia in compound forms like είτε or
μήτε/οΰτε (o - u - q e is Mycenaean) but otherwise was reintroduced from outside as
part of learned language. That είτε etc. belong to a continuous tradition is confirmed
by IG V 2, 262 (Mantinea) where είτε (or rather εϊΗε) is written with tsade, which is
used in parts of Arcadia to indicate the continuation of a labiovelar, i.e an intermediate
stage between a voiceless labiovelar and [t]. This excludes the possibility of a bor­
rowing. Yet in the same Mantinea inscription (line 19, see (2) above) we find τε writ­
ten with tau, while we would expect Με. Could this be a sign that this use of τε is not
pure Arcadian? A counterargument, however, is the Pheneos (?) inscription (1) above)
where we find ... ζ ’· ... κα (for τε ... κάς) and ζ ’ represents again the treatment of a
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 55

labiovdar; here too a borrowing could not have happened in this form. Probably we
cannot reach a firm conclusion, but at least it seems clear that in Arcadian—except for
compounds and possibly for the sequence te ... κάς/καί—the simple τε is no longer
in real use.

3.3. κ ά ς/κ α ί, δέ, μεν

We are now left with three Arcadian particles which we ought to consider together:
κάς/καί, δέ, and μέν. Of these μέν is rare while κάς/καί and δέ are not. The form
κάς, as we have seen, is also Cypriot but in the connective function is not Mycenaean
(as far as as we know); καί is panhellenic—with the exception of Cyprus—and is
general in Arcadia except for the early inscriptions from Mantinea. By the mid-fourth
century Mantinea too has καί. Nowhere do we find alternations between κάς and
καί in the same text. This is not the place where to discuss again the etymology of
κάς/καί,7 but on any hypothesis we need to account a) for the contrast between
Cyprus and Arcadia, b) for the contrast in Arcadia between Mantinea and the other
cities, c) for the contrast between early and late Mantinea. It would be possible of
course to assume that the whole of Arcadia had at some stage both κάς and καί and
each city made its choice but even on this hypothesis we sitali have to assume that
Mantinea first selected κάς and then rejected that choice or rather replaced κάς with
καί. The likelihood is that Mantinea was influenced in its later choice of καί by other
dialects. In other words the καί of Mantinea is a borrowing, conceivably from other
parts of Arcadia or from other parts of Greece. If so, however, one may well wonder
if καί in the rest of Arcadia could not have the same explanation and whether we
should not assume that in the whole of Arcadia the earlier form was κάς (cf. Ruijgh
1981).
As for δέ its absence from Cyprus can only be due to an independent develop­
ment in that dialect, since δέ is well attested in Mycenaean and in Arcadian. Cyprus
itself may preserve traces of δέ possibly in m e - t c and almost certainly in ίδέ, if that
word is correcdy etymologized as containing δέ (Ruijgh 1967: 57).8

Some basic points are now established: a) Arcadian—or at least the Arcadian of the
inscriptions—has only a small number of particles, b) the quasi disappearance of τε
must be due to a reasonably fast evolution in usage after the Mycenaean period, c) the
form καί rather than κάς of Mantinea and possibly of the rest of Arcadia is due to
external influence.

7 The fullest discussion is still that by Liittcl (1981); cf. also the review by Ruijgh (1981).
8 Cypriot also provides good evidence for the deictic pronoun ΰδε. If this derives from the *so
pronoun and connective δέ (see below p. 64), we have further evidence for the earlier use of δέ in
the ancestor of classical Cypriot.
56 ANNA MORPURGO DAVIES

3.4. δε: function and use

It is time now to concentrate on δέ and to move from questions of form to questions

of function and use.
In a paper read at a Dutch Colloquium on Greek particles it is not necessary to
survey the various new approaches to the analysis of particles, since most of the work
done arises in the Netherlands. It is sufficient to refer to the general discussion by
Kroon (1995) about discourse particles in Latin and, more specifically, to Bakker’s
article (1993) on Greek δέ and the earlier monograph by Stephen Levinsohn on tex­
tual connections in A c t s (1987). For epic use in general the most important contribu­
tion is Ruijgh (1971), which also summarizes the Mycenaean data discussed in
Ruijgh (1967); for Arcadian the basic data are collected in Dubois (1986) and some
very relevant observations appear in Hodot (1990).
There are at least two ways to discuss the use of connective particles like καί and
δέ: one may consider what they ‘mean’, their semantic or propositional content, or
one may consider what pragmatic function they have; there may of course be overlap
between the two enterprises. In a discourse context it is also necessary to consider the
unit to which the particles belong. Levinsohn (1987), for instance, discussing the use
of the connective particles in the Acts of the Apostles observes that δέ and very rarely
άλλά, γάρ, οΰν and τότε mark what he calls a development unit and that καί and τε
operate within that development unit. Bakker (1993) has contrasted the oral tradition,
represented e.g. in epic poetry, where δέ may mark a cognitive unitor the equivalent
of an intonation unit, with the written tradition where δέ still marks a unit, but one
which is more content oriented and more deliberate.
So much can be accepted but, when dealing not with literary material but with
inscriptions written in formal style, there are or may be considerable difficulties in
understanding what aims are fulfilled by the particles used. Consider for instance a
straight formula such as that of a late proxeny decree which provides for the award of
proxeny together with consequent honours and perquisites for the honorand. In the
same Arcadian city (Orchomenos) and in the same century (third century B.C.) we
find on the one hand the standard formula in (5) and on the other the formula in (6):

(5) Schwyzer DGE 667, Dubois (1986: ii, 175, no. 11), Thür and Taeuber (1994:
345, Orchomenos, third century):
Έ δοξε ται βωλαι καί τα ι πόλι των Όρχομενίων Λάρχιππον Στιπάκω
Τεγεάταν πρόξενον ην(α]ι καί εύεργέταν ... αύτόν καί έγγόνος- η ν α ι δέ
α ύ τ ώ ι γας ’ίνπασιν καί ατέλειαν καί άσυλί[αν ...
‘It was agreed by the assembly and by the city of Orchomenos that Larchippos
son of Sdpakos from Tegea should become proxenos and benefactor ... he and
his descendants, and (δέ) that he should have land property and freedom from
taxes and inviolability...’
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 57

(6) Dubois (1986: ii, 174, no. 10), Thür and Taeuber (1994: 345, Orchomenos, third
century):

"Εδοξε τα ι π[ό]λι των Όρχομενέων Πάννιν Αίγιπίου Α ργείον πρόξενον

εΐμεν κ α ί εΰεργέτ[α]ν αυτόν [καί γένος κ α ί ά σ υ [ λ ] ία ν ή μ ε ν [α ύ ] τ ώ ι
καί γ έ ν ε ι ...
‘It was agreed by the city of Orchomenos that Pannis son of Aigypios from
Argos should become proxenos and benefactor, he and his family, and (καί) that
he and his family should have inviolability ...’

We have difficulties here in deciding what prompts the use of a Sé clause in (5) and
of a καί clause in (6). It is of course possible that in (6) the choice of καί rather than
Sé is determined by dialect difference, since (6) lias a number of features which are
not Arcadian (είμεν/ημεν, -ου genitives, έν, έπί + genitive) and which have induced
Dubois to state that the dialect is that of the honorand, i.e. Argolic, but this is far from
certain. In a similar proxeny formula from Argos (SchwyzerDGE 92, third century)
we find a Sé construction parallel to that of (5) (πρόξενον ήμεν καί εύεργέταν ...
αυτόν καί έκγόνονς, ήμεν 5έ σφιν ατέλειαν κ αί ασυλίαν .,.). Also in a Mantinea
proxeny decree for an Argive, once again not in Arcadian, we find the standard Sé
construction (IG V 2 263). On the other hand in an earlier proxeny decree from
Tegea possibly of the late fourth century (Dubois 1986: ii, 81, no. 8; Thür and
Taeuber 1994: 342) the usual formula is introduced by δέ καί (ήναι δε καί
[άσυλία]ν κ α ί ατέλειαν).

Examples of this type could be multiplied, but we can also use (5) by itself, as well as
the alternation between δέ and καί in (5) and (6), to underline the point that in some
Arcadian texts δέ simply introduces a new point without any contrapposition; in
other words the correct English rendering is ‘and’ rather than ‘but’.9 In general if
one looks at late Arcadian inscriptions this is the main function of δέ. At its simplest
it marks a sectioning of the discourse indicating that a different piece of information
is now introduced, in fact what Bakker calls a thematic break.

3.5. δ έ in parallel texts

The last conclusion is expected. Yet if we contrast early andiate texts we observe that
initially δέ is used less frequently or more discriminately than later on. An objectively
usable example, even if it raises the problem of different dialects, is that of two paral­
lel texts which define boundaries. The first—from Fourth Century Orchomenos and

9 In (5) it would be possible to argue that the contrapposition is between the title of proxenos and
benefactor awarded to Larchippos and his descendants (αύ-tòv καί έγγόνος) and the rights to
inviolability etc. awarded to Larchippos only (αύτφ); in fact the parallelism with formulae in which
instead of a singular we find a plural pronoun (αύτοις) or a reference to the έκ'/όνοις, while still
having δέ as in (5), show that this is not the critical point (cf. e.g. IG V 2 10, ibid., 17, ibid. 263,
Dubois 1986: ii, 169, no. 6, ibid. p. 177, no. 13, IG V 2 389, ibid., 510, etc.).
58 ANNA MORPURGO DAVIES

i n A r c a d ia n — h a s a s t a n d a r d p a t t e r n ‘f r o m X t o Y ’ w h i c h i s c o n s t a n t l y r e p e a t e d ; n o
δ έ a p p e a r s . T h e s e c o n d is i n k o i n a a n d f r o m P h i g a l i a m o r e th a n o n e c e n t u r y l a te r ; i t
i s b r o k e n b u t i t is c l e a r t h a t t h e p a t t e r n i s s i m i l a r t o t h a t o f O r c h o m e n o s , t h o u g h α π ό
τ ώ ι ν υ ‘f r o m l i d s ’ is r e p l a c e d b y τ ο ύ τ ω δ έ ‘f r o m t h i s ’ a n d ά π ό + n a m e i n th e d a t i v e
i s r e p l a c e d b y ά π ό δ έ + n a m e i n t h e g e n i t i v e . I n o t h e r w o r d s h e r e e a c h c l a u s e is
in t r o d u c e d b y δ έ i n c o n tr a s t w ith th e e a r l i e r te x t:

( 7 ) S c h w y z e r , D G E 6 6 4 , D u b o is ( 1 9 8 6 : i i , 1 1 3 , n o . 1 ), T h ü r a n d T a e u b e r ( 1 9 9 4 : 1 2 4 ,
n o . 1 4 , O r c h o m e n o s , f i r s t p a r t o f t h e F o u r t h C e n tu r y ; s e e T h ü r a n d T a e u b e r 1 9 9 4 :
1 2 5 ). M a r k i n g o f b o u n d a r ie s :

ά π ύ τ ώ ι ό ρ ί ο ι ... έ π ί τ ο Β ο υ φ α γ έ ο ν μ ε σ α κ ό θ ε ν τ ο ΐ ς κ ρ ά ν α ι υ ν ά π ύ τ ώ ι ν ι ί ν
τ ά ν Π ο ρ Θ ι έ α ν π ρ ώ τ α ν ά π ύ τ ώ ι ν ι ... 1 ν η ά τ α ν ά π ύ τ ώ ι ν υ ί ν τ ά ν ά λ ω ν α
ν η ά τ α ν ... ά π ύ Π α δ ο έ σ σ α ι ί ν τ ο ν λ ό φ ο ν τ ο ν δ ί ω ρ ο ν κ α ί ά π ύ τ ώ ι ν υ ί ν τ ά ν
Τ ρ ι ά ν γ κ ε ι α ν ά π ύ τ ώ ι ν υ ...
‘f r o m th e b o u n d a r y ... to th e B o u p h a g e o n ( p a s s in g ) i n t h e m id d le o f th e s p r in g s ;
f r o m th is p o i n t t o t h e b e g i n n i n g o f t h e P o r th ie a ; f r o m th is p o i n t t o th e e n d ( o f th e
P o r th ie a ) ; f r o m t h i s p o i n t t o th e e n d o f th e o r c h a r d ; ... f r o m th e P a d o e s s a t o th e
h ill w ith t w o b o u n d a r ie s , a n d f r o m th is p o i n t t o th e T r ia n k e ia ; f r o m th is p o i n t . . . ’.

( 8 ) D u b o is ( 1 9 8 6 : i i , 2 6 6 , n o . 3 ); T h ü r a n d T a e u b e r ( 1 9 9 4 : 3 0 1 , n o . 2 9 , P h i g a l i a , s e c ­
o n d p a r t o f th e T h i r d C e n tu r y ; in k o in a ) . M a r k i n g o f b o u n d a r i e s ( v e r y f r a g m e n ta r y ) :

... τ ο ύ τ ω δ έ έ [ π ] ί [ . . . ά π ό δ [ έ ... τ ο ύ τ ω δ ε ... , 101

I n s p ite o f th e d i f f e r e n t d ia l e c t t h e c h r o n o l o g i c a l c o n tr a s t is r e p r e s e n ta tiv e . M o r e g e n ­
e r a lly th e c o n t r a s t b e t w e e n e a r ly a n d l a t e is a c o n t r a s t b e t w e e n li m i t e d u s e o f δ έ a n d
e x te n s iv e u s e o f δ έ .

3.6. δέ in some longer Arcadian inscriptions

S o m e f i f t h a n d e a r l y f o u r th c e n t u r y i n s c r i p t i o n s s h o w s u b t l e r d i s t i n c t i o n s i n t h e u s e
o f δ έ . 11 F r o m t h i s p o i n t o f v i e w t h e l o n g e r te x ts i n A r c a d i a n f a l l i n t o t w o g r o u p s .
T h e s m a l l e r g r o u p i n c l u d e s t h e L a w a b o u t th e c u l t o f D e m e t r a T h e s m o p h o r o s
( P h e n e o s ? , c a . 5 0 0 B . C . ) , 12 th e j u d g e m e n t a b o u t th e m u r d e r i n th e t e m p le o f A l e a
( M a n t i n e a , F i f t h C e n t u r y B . C . ) 13 a n d t h e r e g u l a t i o n s f o r t h e t e m p l e o f A t h e n a A l e a

10 For a similar set o f formulae cf. Thür and Taeuber (1994: 306, no. 3 1 ,1 A (IG V 2 p. xxvii)), a
second century fragmentary inscription from Olympia with a boundary division with Megalopolis.
11 Hodot (1990) offers some subtle analyses of the use of δέ in a few inscriptions. However, his
aim (viz. to establish the existence or otherwise of the potential particles καν and δαν in Arcadian)
is different from mine and I hope to extend the material while looking at it from a somewhat
different angle.
12 Dubois (1986: ii, 196, no. 1), Thür and Taeuber (1994:269. no. 20; cf. (1)).
13 IG V 2 262, Schwyzer DGE 661, Dubois (1986: ii, 94), Thilr and Taeuber (1994: 75 no. 8; cf.
( 2 ))-
 PARTICLES IN’GREEK EPIGRAPHICAL TEXTS 59

( T e g e a , e n d F i f t h C e n t u r y o r b e g i n n i n g F o u r t h C e n t u r y ) . 14 T h e s e c o n d g r o u p
i n c l u d e s m o s t o f th e r e m a i n i n g d i a l e c t in s c r i p t i o n s f r o m A r c a d i a o f s u i t a b l e le n g th .
P r o m i n e n t e x a m p l e s a r e th e r u l e s a b o u t b u il d i n g c o n t r a c t s f r o m T e g e a ( S e c o n d h a l f
o f th e F o u r t h C e n t u r y ) , 15 th e d e c r e e f o r th e r e t u r n o f th e e x ile s a ls o f r o m T e g e a ( 3 2 4
B . C . ) , 16 t h e m id - f o u r t h c e n t u r y b o u n d a r y s t a t e m e n t f r o m O r c h o m e n o s m e n t i o n e d
a b o v e ( i n ( 7 ) ) , a n d t h e v a r io u s p r o x e n y d e c r e e s i n d i a l e c t ( s e e a b o v e ) . T o t h i s g r o u p
m a y a l s o b e l o n g th e s y n o ik ia b e t w e e n O r c h o m e n o s a n d E u a i m o n ( O r c h o m e n o s , m id
f o u r t h c e n t u r y ) 17 a n d th e s y n o i k i a b e t w e e n t h e H e l i s w a s i o i a n d th e M a n t i n e a n s
( M a n t i n e a , d a t e u n c e r t a i n b u t a t l a t e s t 3 5 0 - 4 0 a n d p o s s i b l y e a r l i e r ) , 18 t h o u g h th e s e
t w o i n s c r i p t i o n s a n d p a r t i c u l a r l y th e l a t t e r s e e m t o h a v e a n i n t e r m e d i a t e p o s i t i o n
b e tw e e n t h e t w o g r o u p s . I t i s n o t i c e a b l e t h a t t h e t w o g r o u p s a l s o d if f e r o n c h r o n o l o g ­
i c a l g r o u n d s , w ith t h e f i r s t g r o u p in c l u d i n g a l l th e e a r l y in s c r ip tio n s .
F u n d a m e n t a l l y δ έ is a c o n n e c t i v e p a r t i c l e w h i c h l i n k s s e n t e n c e s r a t h e r th a n
c l a u s e s . A s s u c h i t n o r m a l l y ta k e s s e c o n d p o s i t i o n i n th e f i r s t c l a u s e o f t h e s e n t e n c e
a f te r th e f i r s t a c c e n te d w o r d . B e c a u s e δ έ is a c o n n e c tiv e w e d o n o t e x p e c t i t t o a p p e a r
i n t h e f i r s t s e n t e n c e o f a t e x t a n d in d e e d i t n e v e r d o e s . M o r e d e l i c a t e i s t h e q u e s t i o n
o f th e s e c t i o n s i n w h i c h a t e x t m a y b e d iv id e d . T h e c l e a r e s t i n s t a n c e s a r e t h o s e o f
in s c r ip tio n s w h e r e th e d iv is io n in to p a r a g r a p h s is g r a p h ic a lly m a r k e d e ith e r by
d o u b l e p u n c t u a t i o n s (a s i n t h e r e g u l a t i o n s f o r th e t e m p l e o f A t h e n a A l e a ) o r b y a n
h o r i z o n ta l l i n e (a s i n t h e s y n o i k i a o f th e F l e lis w a s io i a n d th e M a n tin e a n s ) o r b y a l in e
a n d a n e m p t y l e t t e r s p a c e (a s i n th e r u l e s a b o u t b u i l d i n g c o n tr a c ts ) . N o t i c e a b l y i n th e
e a r lie s t o f th e s e te x ts ( th e f ir s t) n o p a r a g r a p h h a s a n in i t i a l δ έ ; t h e s e c o n d t e x t f o llo w s
th e s a m e r u l e w i t h o n e p o s s i b l e e x c e p t i o n . 19 T h e t h i r d te x t, o n th e c o n tr a r y , r e g u la r ly
b r e a k s th e r u l e w h i l e a t t h e s a m e t i m e a ls o b r e a k in g ( o n c e a t l e a s t) t h e r u l e a c c o r d in g
to w h i c h δ έ is i n s e r t e d a f t e r th e f i r s t w o r d o f th e s e n t e n c e ( I G V 2 6 , 2 1 : Μ ή έ ξ έ σ τ ω
δ έ ...). I f w e l o o k a t o t h e r i n s c r i p t i o n s w h e r e s e p a r a t e p a r a g r a p h s o r s e c t i o n s c a n
o n l y b e i d e n t i f i e d o n s e m a n t i c b a s e s w e o b t a i n s i m i l a r r e s u l t s . I n th e f i f t h c e n t u r y
m u r d e r j u d g e m e n t f r o m M a n t i n e a a s e r io u s c h a n g e o f to p ic c a lls f o r t h e a b s e n c e o f
i n i t i a l δ έ ; 20 b y c o n t r a s t i n t h e m u c h l a t e r T e g e a d e c r e e f o r th e r e t u r n o f e x i l e s δ έ
s e n t e n c e f o l l o w s δ έ s e n t e n c e s o th a t, i f th e p r e v i o u s c r i t e r i o n w e r e t o b e a d o p te d ,

14 IG V 2 3; Schwyzer DGE 654, Dubois (1986: Si, 20), Thür and Taeuber (1994: 11 no. 2).
15 IG V 2 6; Schwyzer DGE 656, Dubois (1986: ii, 39), Thür and Taeuber (1994: 20, no. 3).
16 IG V 2 p. xxxvi; Schwyzer DGE 657, Dubois (1986: ii, 61 no. 4), Thür and Taeuber (1994: 51
no. 5).
17 IG V 2 343, Schwyzer DGE 665, Dubois (1986: ii, 146), Thür and Taeuber (1994: 130 no. 15).
18 Te Riele (1987), Thür and Taeuber (1994: 98 no. 9).
19 After the division sign the new paragraph at line 13 starts with Ό σ α δέ συνβόλ[α]ια
έτύγχανον έχο ντες ο ι Έ λ ισ ^ά σ ιο ι αυτοί πό αυτός πάρος Μαντινής έγένοντυ ... 'However
many contracts the Heliswasioi had among themselves before they became Mantineans, will be valid
The unique presence of δέ in this paragraph may raise the question whether we should not read
Ό σάδε, accepting the existence of a univerbated όσόσδε parallel to τοσόσδε, which, as shown by
Risch (1969), arose perhaps in the post-Myccnacan period, but panhellenically, from an earlier
sequence τόσος δέ.
20 Notice the absence of δέ in lines 14,18,30. The only possible exception is in line 24 where the
editors restore Εύχολα [δ1] όδε έσετοι το ί ά[λιτερίοι, but it is difficult to build loo much on a
restoration.
60 ANNA MORPURGO DAVIES

m o s t o f th e i n s c r i p t i o n w o u l d c o u n t a s o n e im m e n s e p a r a g r a p h . T h i s m a y b e i n d e e d
t h e c o r r e c t i n t e r p r e t a t i o n i f o n e th in k s o f t h e in v o l u t e d l e g a l s t y l e w h i c h p r e v a i l s i n
l a t e r te x ts — b u t o b v i o u s l y t h e r e is a c e r t a i n e l e m e n t o f c i r c u l a r i t y i n t r y i n g t o a r g u e
t h a t n e w p a r a g r a p h s n e v e r s ta r t w i t h δ έ , w h i l e t h e n a r g u i n g t h a t i n s o m e in s c r i p t i o n s
p a ra g ra p h s a re v e r y lo n g b e c a u s e m o s t s e n te n c e s s ta rt w ith δ έ . M o re o v e r th e p a ra l­
le l i s m w i t h t h e b u i l d i n g i n s c r ip tio n w h e r e p a r a g r a p h s a r e g r a p h i c a l l y m a r k e d s p e a k s
f o r th e i n t r o d u c t i o n o f δ έ a t th e b e g i n n i n g o f p a r a g r a p h s . T h e tr u th is p r o b a b l y i n th e
m id d le . W i t h i n c r e a s e d s o p h is tic a tio n i n ‘le g a l e s e ’ i t b e c o m e s m o r e a n d m o r e n e c e s ­
s a r y t o m a r k f o r m a l l y th e l in k s b e t w e e n t h e d i f f e r e n t p a r t s o f a t e x t in c l u d i n g p a r t s
w h i c h c o n c e r n d i f f e r e n t s u b je c ts a n d w h i c h c a n c o n s e q u e n t l y b e m a r k e d a s b e l o n g ­
i n g to d if f e r e n t p a r a g r a p h s .

3.7. δ έ in texts of different periods

E v e n o n th i s s i m p l e t e s t w h ic h c o n c e r n s th e p r e s e n c e o r a b s e n c e o f δ έ a t t h e b e g i n ­
n in g o f p a r a g r a p h s th e e x is te n c e o f a lin g u is tic c o n tr a s t b e tw e e n e a r ly a n d la te
in s c r ip tio n s is c o n f i r m e d , b u t th e p h e n o m e n o n i s u n d e r l i n e d i f w e a c tu a lly l o o k a t th e
m e a n i n g o f δ έ i n te x ts o f d i f f e r e n t p e r i o d s . I n th e le x s a c r a a b o u t th e c u l t o f D e m e t r a
T h e s m o p h o r o s , i n t h e ju d g e m e n t o f M a n t i n e a a n d i n th e d i s p o s i t i o n s f o r th e t e m p l e
o f A th e n a A le a , δ έ i s r e g u l a r l y u s e d to i n d i c a t e a c o n t r a s t w i t h t h e p r e v io u s s e n te n c e .
I n p r a c tic a lly a l l in s t a n c e s th e c o r r e c t tr a n s l a t i o n is ‘b u t ’ r a t h e r t h a n ‘a n d ’; t h a t t h i s is
s o is s h o w n b y t h e f r e q u e n c y o f δ έ c l a u s e s w h i c h n e g a t e t h e p r e v i o u s c la u s e , o f t e n
e llip tic a lly , b u t s o m e tim e s w ith a n a c tu a l r e p e titio n o f th e v e r b :

( 9 ) R e g u la tio n s f o r th e te m p le o f A th e n a A l e a (n o te 1 4 ), 1 .1 5
# T à h i e p à π ρ ό β α τ α μ ε ν έ μ ε ν ί ν Ά λ έ α ι π λ ό ς ά μ έ ρ α υ κ α ν ν υ κ τ ό ς ... ■ ε ί δ ’ α ν
ν έ μ ε ...
‘T h e s a c r e d c a t t l e m u s t n o t p a s t u r e i n A l e a m o r e th a n a d a y a n i g h t ... b u t ( δ έ ) i f
it p a s tu r e s ...’

( 1 0 ) i b id ., 1 .5

Τ ο ν Ιη ερ ο θ ύ τ α ν ν έμ ε ν ίν Ά λ έ α ι ο τ ι αν ά σ κ ε θ έ ς ε , τά δ’ άνασκεθέα
ίν φ ο ρ β ίε ν
‘L e t th e s a c r i f i c e r p a s t u r e i n A l e a w h a t e v e r i s u n b l e m i s h e d , b u t ( δ έ ) f o r t h e a n i­
m a ls w h i c h a r e n o t u n b le m is h e d l e t th e r e b e inphorbismos’

( 1 1 ) L a w a b o u t t h e c u l t o f D e m e t r a T h e s m o p h o r o s ( n o te 1 2 ), 1. 4

... ο ζ ι ς τ ό τ ε δ α μ ι ο ^ ο ρ γ έ ά φ ά ε σ τ α ι δ α ρ χ μ ά ς τ ρ ι ά κ ο ν τ α ' ε ί δ έ μ έ ά φ ά ε τ ο ι . . .
‘w h o e v e r is t h e n a m a g is tr a te w i l l p a y t h ir ty d r a c h m a i, b u t ( δ έ ) i f h e d o c s n o t p a y
 PARTICLES IN GREEK EPIGRAPfflCAL TEXTS 61

I n th e s e t e x t s δ έ d o e s n o t s i m p l y s e c t i o n t h e d i s c o u r s e w h i l e h i g h l i g h t i n g i t s c o n t i ­
n u i t y b u t a c t u a l l y a d d s a s e m a n t i c c o m p o n e n t . 21 I n th e l a t e r t e x ts th e p o s i t i o n i s d if ­
f e r e n t; p r a c t i c a l l y e v e r y n e w s e n t e n c e is li n k e d t o t h e p r e c e d i n g o n e b y δ έ w h i c h h a s
a c o n t i n u a t i v e r a t h e r th a n a d v e r s a t i v e v a lu e , T h e c o n t r a s t is b e s t il l u s t r a t e d b y f o r m u ­
l a e w i t h s i m i l a r f u n c t i o n w h i c h b e l o n g t o th e e a r l y a n d t h e la t e r p e r i o d r e s p e c t i v e l y .
C o n t r a s t o n c e a g a i n th e r u le s f o r th e a d m i n is tr a tio n o f th e te m p le o f A t h e n a A l e a a n d
th e s o m e w h a t la t e r b u ild in g te x t, a l s o f r o m T e g e a :

(1 2 ) R e g u l a t i o n s f o r th e t e m p le o f A th e n a A l e a ( n o te 1 4 ) 1 .1 ff.:

# x ò v h ie p è v π έ ν τ ε κ α ι ε ί κ ο σ ι ο ι ς ν έ μ ε ν κ α ι ζ ε ύ γ ο ς κ α ί α ίγ α ·
ε ί 6' α ν κ α τα λ λ ά σ σ ε
ίν φ ο ρ β ισ μ ό ν έ ν α ν
τ ο ν Ι ιιε ρ ο μ ν ά μ ο ν α ίν φ ο ρ β ίε ν .
ε ί δ ’ α ν λ ε ύ τ ο ν μ έ ίν φ ο ρ β ίε
Ι ιε κ ο τ ό ν δ α ρ χ μ ά ς ό φ λ ε ν ί ν δ α μ ο ν
κ α ί x ó ta p fo v S v ai
‘T h e p r i e s t m a y p a s tu r e tw e n t y f i v e s h e e p , a y o k e a n d a g o a t;
b u t ( δ έ ) i f h e g o e s b e y o n d th is ,
le t th e re b e inpkorbismos·,
l e t th e h i e r o m n a m o n p r a c tic e th e inpkorbismos.
b u t ( δ έ ) if , h a v in g s e e n it, h e d o e s n o t p r a c t i c e in p h o r b is m o s
l e t h im o w e o n e h u n d r e d d r a k h m a i t o th e p e o p le
a n d (κ α ί) be a c c u rs e d 1

(1 3 ) R u l e s a b o u t b u ild in g c o n t r a c t s ( n o te 1 5 ), 1. 2 1 ff .

# Μ ή έ ξ έ σ τ ω δ έ μ η δ έ κ ο ιν α ν α ς γ ε ν έ σ θ α ι π λ έ ο ν ή δύ ο έ π ί μ η δ έ ν ι τω ν
έργω ν-
ε ί δ έ μή,
όφ λ έτω έ κ α σ το ς π ε ν τή κ ο ν τα δ α ρ χ μ ά ς'
έ π ε λ α σ ά σ θ ω ν δ έ ο ί ά λ ια σ τ α ί-
ί μ φ α ί ν ε ν δ έ τ ό μ β ο λ ό μ ε ν ο ν έ π ί τ ο ΐ ή μ ί σ σ ο ι τ α ς ζ α μ ί α υ ...
‘L e t i t n o t b e a llo w e d t o h a v e m o r e t h a n tw o p a r tn e r s f o r a n y w o r k
b u t ( δ έ ) i f n o t,
e a c h ( o f th e p a r tn e r s ) m u s t p a y f if ty d r a c h m a i,
a n d ( δ έ ) th e h e l i a s t a i w i l l im p o s e ( th e f in e ) ;
a n d ( δ έ ) w h o e v e r w i s h e s c a n d e n o u n c e ( th e m ) f o r h a l f o f t h e f i n e ’

2 1 1 deliberately leave aside at this stage the μεν ... δέ constructions, which will be discussed later
(§ 7).
62 ANNA MORPURGO DAVIES

I n b o t h t e x ts t h e f i r s t h y p o t h e t i c a l s e n t e n c e is f o l l o w e d b y a s t a t e m e n t a b o u t th e
m a g is tr a te s i n c h a r g e o f th e im p o s i t i o n o f t h e p e n a lty ; i n t h e f i r s t te x t th is is i n a s y n ­
d e t o n , i n t h e s e c o n d ( a n d la te r te x ts ) i t i s in t r o d u c e d b y δ έ . 22
I n g e n e r a l i t a l l l o o k s a s i f i n A r c a d i a n i n s c r ip tio n s th e u s e o f δ ε c h a n g e d r o u n d
t h e m id d le o f th e f o u r t h c e n tu r y . A t a l l p e r i o d s th e p a r t i c l e a c ts a s a s e n te n c e c o n n e c ­
tiv e ,23 b u t a t a l a t e r s t a g e i t c a n a p p e a r a t t h e b e g in n in g o f m o s t s e n te n c e s m a r k i n g o n
t h e o n e h a n d th e c o h e s i o n w i t h i n a te x t, o n th e o t h e r th e m a t i c a n d g r a m m a tic a l s h if ts
o f a l l ty p e s . P r e v i o u s l y th i s w a s n o t s o a n d δ έ r e q u i r e d a m o r e s e m a n t i c a l l y b a s e d
d e f in itio n ; i t in t r o d u c e d a r e b u t t a l o f p r e v i o u s s ta te m e n ts , i. e . w a s s e r io u s ly a d v e r s a ­
tiv e r a th e r th a n m e r e ly c o n tin u a tiv e o r tr a n s itiv e .

4. From adversative to transitive Sé?

O b v i o u s l y t h e c o n t r a s t j u s t i d e n t i f i e d m a y b e th e d u e t o t h e l i m i t s o f o u r e v i d e n c e
s i n c e i n t h e e a r l y p e r i o d th is is e x i g u o u s . Y e t t h e p a t t e m t h a t w e h a v e f o u n d is
re m a rk a b ly c o n s is te n t a n d a t p r e s e n t w e m u s t a c c e p t it as a sig n ific a n t. I f so , a n d i f
t h e r e is a c o n t r a s t b e t w e e n A r c a d i a n a n d o t h e r f o r m s o f G r e e k a s w e ll a s a d e v e l o p ­
m e n t w ith in A r c a d i a n , w e o u g h t t o tr y t o u n d e r s t a n d h o w t h i s c a m e a b o u t. A n o b v i ­
o u s h y p o t h e s i s is t h a t t h e e a r l i e s t t e x t s r e p r e s e n t t h e o r i g i n a l a d v e r s a t i v e a n d q u a s i
a d v e r b i a l m e a n i n g o f δ έ w h ic h l a t e r o n w a s w e a k e n e d a l l o w i n g a m u c h w i d e r r a n g e
o f u s e s . I f s o , th is w o u l d p r o v i d e m u c h n e e d e d s u p p o r t f o r t h e f i r s t o f th e t w o c o n ­
t r a s t i n g v i e w s a b o u t th e o r i g i n o f δ έ . B a k k e r ( 1 9 9 3 ) h a s h i g h l i g h t e d t h e c o n t r a s t
b e tw e e n t h o s e ( l i k e K ü h n e r - G e r t h 2 , 2 6 1 f f .) w h o b e lie v e t h a t t h e o r ig in a l f u n c t i o n o f
t h e p a r t i c l e is a d v e r s a t i v e ( ‘b u t ’, ‘h o w e v e r ’) a n d th o s e , l i k e C J . R u i j g h ( 1 9 7 1 :
1 2 8 f f .) , w h o a r g u e t h a t t h e m a in p u r p o s e o f th e p a r t i c l e i s to i n d i c a t e tr a n s itio n f r o m
o n e p o i n t o f th e n a r r a t i v e to a n o th e r . A t r a n s i t i v e v a l u e , a s w e h a v e s e e n , d o e s n o t
a c c o u n t f o r th e u s e o f δ έ i n e a r ly A r c a d i a n , a n a d v e r s a tiv e v a l u e d o c s . F o r th e f o l l o w ­
e r s o f th e K ü h n e r - G e r t h t h e o r y th e n A r c a d i a n m u s t b e s i n g u l a r l y a r c h a ic . Y e t i f w e
t a k e th is p o i n t s e r i o u s l y w e m u s t a l s o a r g u e t h a t o u r in s c r i p t i o n s a r e m o r e a r c h a i c i n
th e ir u s e o f δ έ th a n H o m e r a n d H e r o d o tu s , a n d p e rh a p s e v e n M y c e n a e a n — w h ic h
d e f ie s c r e d ib ility . T h e la s t p o in t m u s t b e e x p l o r e d f u r th e r .
I n M y c e n a e a n δ έ is c e r ta in ly u s e d i n c o n t r a p p o s itio n s . I t is e n o u g h to r e m e m b e r
i n t h e f i r s t in s t a n c e t h e c o n t r a s t b e t w e e n t h e d i f f e r e n t s t a t u s o f f a t h e r a n d m o t h e r i n
th e do-qe-ja i n s c r ip tio n :

(1 4 ) P Y A n 6 0 7 , 5 f f .:

... do-e-ropa-te

22 In the second text I have always rendered δέ with either 'and' or 'but' to show that a 'but'
meaning is not always appropriate or possible; obviously in a real translation other renderings
might be more appropriate.
23 This is the prim ary function, which explains, for instance, why δέ only appears in the first
component of a hypothetical sentence; a construction like ε ί δέ μή, δαρχμάν <5φλέν is grammatical,
but a construction like “ δαρχμάν όφλέν, ε ί δ έ μή is not.
 PARTICLES IN GREEK EPIGRAPfflCAL TEXTS 63

ma-te-de di-wi-ja do-c-ra

... do-e-ra ma-te
pa-te-de ka-ke-u
‘... th e f a t h e r ( is ) a s la v e , b u t (de) th e m o t h e r ( is ) a s la v e o f D iw ia ;
.... t h e m o t h e r ( is ) a s la v e , b u t (de) t h e f a th e r (is ) a b r o n z e s m i t h ’

A n e q u a l l y c l e a r e x a m p l e i s th e f a m o u s c o n t e s t b e t w e e n t h e p r i e s t e s s a n d th e c o m ­
m u n ity :

(1 5 ) P Y E p 7 0 4 , 5 -6 :

e-ri-ta i-je-reja e-ke

e-u-ke-to-qe e-to-ni-jo e-ke-e te-o
da-mo-de-mi pa-si ko-to-na-o ke-ke-me-na-o o-na-to e-ke-e
Έ . th e p r ie s te s s h as
and (qe) s o le m n ly a f f ir m s t h a t s h e h a s th e e-to-ni-jo f o r th e g o d (d e ss)
but (de) th e damos s a y s t h a t s h e h a s a n onato o f ktoinai kek'

A s i m i l a r m e a n i n g is f o u n d in th e t e x t w ith w h i c h R u i j g h ( 1 9 6 7 : 3 3 7 ) s t a r ts h i s a n a l ­
y s is o f M y c e n a e a n δ έ :

(1 6 ) P Y E b 3 3 8 ,1 -2 ;

ka-pa-ti-ja ... e-ke-qe ke-ke-me-no ko-to-[no] dwo

o-pe-ro-sa-de wo-zo-e o-wo-ze
K . ... h a s tw o ktoinai kek.
b u t (de) h a v i n g t o worzeen s h e d o e s n o t worzei’

Y e t n o t a ll M y c e n a e a n in s ta n c e s a re so c le a r-c u t a n d , a s R u ijg h (1 9 6 7 ) h a s s h o w n ,
so m e u ses o f de i n c o n n e c tio n w ith to-so o r in th e p a r tic le s e q u e n c e o-da-a2 s p e a k
f o r ‘t r a n s i t i v i t y ’ i n h is s e n s e r a t h e r th a n f o r a n a d v e r s a t i v e f u n c t i o n . I n a n u m b e r o f
c la u s e s to-so-de i s n o t as y e t a d e m o n s tra tiv e b u t a s e q u e n c e o f τ ό σ (σ )ο ς a n d δ έ , a n d
t h e p a r t i c l e c a n n o t e a s i l y b e u n d e r s t o o d a s ‘b u t ’. T h u s i n t h e P y l o s t a b l e t a b o u t th e
wanakteron temenos t h e e a s i e s t in te r p r e ta tio n i s t h a t in d i c a t e d b e lo w :

(1 7 ) P Y E r 3 1 2 , 5 -6 :

te-re-ta-o toso pe-ma G R A 3 0

to-so-de te-reta V I R 3
So m uch (tos(s)on) s e e d o f th e telestai 3 0 u n its
a n d (de) s o m a n y (tos(s)oi) telestai 3 M e n ’

S im ila r ly in :
 ANNA MORPURGO DAVIES

(1 8 ) P Y Jn 3 8 9 1 ,9 ,1 1 :

a-ka-si-jo-ne ka-ke-we ta-ra-si-ja e-ko-te

to-so-de ka-ko A E S M 27

to-so-de a-ta-ra-si-jo ka-ke-we

‘I n A . b r o n z e s m ith s w h o h a v e(alansian ....

and (de) t h e b r o n z e (is ) s o m u c h : ...
and (de) s o m a n y b r o n z e s m ith s w i t h o u t talansia’

I t i s i n f a c t th is c o n n e c t i v e a n d tr a n s itiv e v a l u e o f δ έ w h i c h o f f e r s th e m o s t p l a u s i b l e
s t a r t f o r th e p r o c e s s e s o f g r a m m a tic a liz a tio n w h ic h le d — d u r i n g , o r p o s s ib ly a f te r , th e
M y c e n a e a n p e r i o d — t o th e c r e a t i o n o f f o r m s l i k e ο δ ε o r τ ο σ ό σ δ ε . R i s c h ( 1 9 6 9 ) ,
w h o a r g u e d f o r a d e r i v a t i o n o f th e ‘d e m o n s t r a t i v e ’ δ έ o f ο δ ε e tc . f r o m th e c o n n e c tiv e
p a r tic le , d id n o t r e a lly a s k h o w it w o u ld h a v e b e e n p o s s ib le to m o v e fro m a p u r e ly
a d v e rs a tiv e v a lu e to a d e m o n s tra tiv e v a lu e ; y e t, i f th e q u e s tio n is a sk e d , th e a n s w e r
m u s t b e t h a t t h e d e v e l o p m e n t is p o s s i b l e i f th e a d v e r s a t i v e v a l u e i s n o t t o o c l e a r l y
m a r k e d . I f , a s R i s c h a r g u e d , th e s h i f t i s h a p p e n i n g i n M y c e n a e a n i ts e lf , i t th e n f o l ­
l o w s t h a t a t t h a t s t a g e δ έ c o u l d n o t b e p u r e l y a d v e r s a tiv e . R a t h e r th a n r e w r i t i n g th e
in te rp re ta tio n o f M y c e n a e a n sub specie Arcadiae, i t w o u l d b e e a s i e r to a s s u m e t h a t
t h e s tr o n g a d v e r s a t i v e v a l u e o f A r c a d i a n δ έ is d u e t o a p o s t - M y c e n a e a n d e l e v e l o p -
m e n t a n d t o t r y t o u n d e r s t a n d h o w th i s c a m e a b o u t. O n e s t a r t i n g p o i n t is t h a t i n o u r
t e x t s t h e r e is n o t a t w o - w a y c o n t r a s t b e t w e e n κ α ί a n d δ έ b u t t h e r e i s r a t h e r a t h r e e -
w a y c o n t r a s t b e w e e n κ α ί , δ έ a n d a s y n d e t o n o r p a u s e . T h e l a s t o f th e s e th r e e h a s t w o
r o l e s . F i r s t i t d e f i n e s l a r g e d e v e l o p m e n t u n i t s , t o u s e L e v i n s o h n ’s t e r m i n o l o g y ,
in c l u d i n g th e w h o l e t e x t o r a p a r a g r a p h . T h e s e a r e t h e u n i t s w i t h i n w h o s e s c o p e δ έ
o p e r a t e s d e f in in g f u r t h e r a n d s m a l l e r u n i t s . I n i t s tu r n i n e a r l y A r c a d i a n κ α ί o p e r a te s
w i t h i n th e s c o p e o f δ έ , s o t h a t t h e r e is a s o r t o f h ie r a r c h y : a s y n d e t o n > δ έ > κ α ί .
S e c o n d l y , a s y n d e t o n c a n b e u s e d w i t h p u r e l y l o c a l v a l u e w i t h i n a u n i t ( e .g . a p a r a ­
g r a p h ) a ls o m a r k e d b y a s y n d e to n ; t h e l o c a l a s y n d e to n s till i n d i c a t e s a n e w th e m e b u t
i n a n e u t r a l f a s h io n . T h i s is w h a t o c c u r s i n ( 1 2 ) , w h e r e , a s w e h a v e s e e n , th e c l a u s u l a
a b o u t t h e i m p o s i t i o n o f t h e f i n e b y t h e l o c a l m a g i s t r a t e f o l l o w s i n a s y n d e t o n th e
p r e v i o u s h y p o th e tic a l s e n te n c e .
W i t h i n th e p a r a g r a p h , th e n , δ έ , a s y n d e t o n a n d s o m e t i m e s κ ά ς / κ α ί a r e a l l p o s s i ­
b l e . O f th e s e κ ά ς / κ α ί , w h i c h is t h e m a i n a n d p r a c t i c a l l y o n l y p a r t i c l e w h i c h j o i n s
n o u n s a n d a d je c tiv e s , in th e A r c a d ia n in s c r ip tio n s is u s e d le s s f re q u e n tly to j o in
c la u s e s . W h e n i t is , i t n o r m a l l y j o i n s c l a u s e s w h i c h s h a r e t h e s a m e s u b j e c t a n d / o r a r e
c l o s e l y l i n k e d f r o m a s e m a n t i c p o i n t o f v i e w . A n e x a m p l e is p r o v i d e d b y th e l a s t
c l a u s e o f (1 2 ) . T h i s m e a n s t h a t a n y o t h e r s o r t o f c o n n e c t i o n a t t h e l o c a l l e v e l w i t h i n
t h e p a r a g r a p h h a d t o b e i n d i c a t e d b y e i t h e r a s y n d e t o n o r t h e u s e o f δ έ . O n th e
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 65

a s s u m p tio n t h a t t h e o r i g i n a l s t a t e o f a f f a ir s w a s s i m i l a r t o t h a t o f M y c e n a e a n w h e r e
δ έ h a s e i t h e r t r a n s i t i v e o r a d v e r s a t i v e v a l u e i t is n o t t o o d i f f i c u l t to s e e w h y i n th e
e a r l y A r c a d i a n in s c r i p t i o n s , w h e r e δ έ c o n t r a s t e d w i t h a s y n d e t o n , i t a c q u i r e d a m o r e
c l e a r l y p o l a r i z e d a d v e r s a t i v e v a l u e , w h i l e a s y n d e t o n s i m p l y i n d i c a t e d t h e s h i f t to
a n o th e r t o p i c w i t h o u t a n y a d d e d s e m a n tic v a lu e .

5. The loss of τ ε and δέ in Cypriot

T h a t th e in te r p r e ta tio n p r o p o s e d is p re fe ra b le to th e a s s u m p tio n th a t e a r ly A rc a d ia n
s i m p l y p r e s e r v e s th e o r i g i n a l m e a n i n g o f δ έ 24 is p e r h a p s s h o w n b y a n a t t e m p t to
m a k e s e n s e o f t h e h i s t o r y o f C y p r i o t w i t h i t s lo s s o f b o t h τ ε a n d δ έ . I f w e a s s u m e
th a t M y c . de h a d a tr a n s itiv e f u n c tio n , i.c . m a r k e d a t h e m a tic s h if t w h ile a ls o id e n t i f y ­
i n g a s e c tio n o f d is c o u r s e , w e p r o b a b l y w a n t t o a t t r i b u t e t o de a b r o a d e r s c o p e th a n to
qe, th e b a s i c c o n n e c t i v e . Y e t t h e c r e a t i o n o f κ ά ς / κ α ί i n th e f u n c tio n o f c o n n e c t o r , o r
r a th e r th e a ttr ib u tio n o f c o n n e c tiv e f u n c tio n s to κ ά ς o r κ α ί , w h ic h m u s t b e p o s t-
M y c e n a e a n , w i l l h a v e a l te r e d t h e s i t u a tio n , p a r t i c u l a r l y b e c a u s e κ ά ς / κ α ί , d i f f e r e n t l y
fro m qe/χε, w a s n o t e n c litic . I t i s g e n e r a l l y a s s u m e d , a n d c o r r e c t l y s o , t h a t κ ά ς / κ α ί
m u s t h a v e ta k e n o v e r th e f u n c tio n s o f qehe. l a r g e l y b e c a u s e i t w a s a ‘s t r o n g e r ’ a n d
‘m o r e e m p h a t i c ’ p a r tic le . T h is , j o i n e d w ith its in i t i a l q u a s i - a d v e r b i a l a n d n o n - e n c litic
n a tu r e , n a t u r a l l y g a v e κ ά ς / κ α ί a r o l e a t t h e b e g i n n i n g o f a s e n te n c e a n d b r o u g h t i t in
s o m e w a y in c o llis io n w ith δ έ . L e t u s a ss u m e — p u r e ly o n a s p e c u la tiv e b a s is — th a t
th is h a p p e n e d i n a n e a r l y p o s t - M y c e n a e a n p e r i o d . A p r i o r i t h e c o n f l i c t b e t w e e n
κ ά ς / κ α ί o n t h e o n e h a n d a n d δ έ o n th e o t h e r c o u l d b e r e s o l v e d i n o n e o f tw o w a y s :
e i t h e r th e r o l e o f b o t h p a r t i c l e s w a s f u r t h e r d e f i n e d a n d p e r h a p s p o l a r i z e d , o r o n e o f
th e t w o p a r t i c l e s w a s g iv e n u p w h i l e th e o t h e r t o o k o v e r it s f u n c tio n s . C o n s i d e r n o w
s o m e o f t h e c o n s t r u c tio n s t h a t w e f i n d i n th e E d a lio n in s c r ip tio n .

(1 9 ) B r o n z e T a b l e o f I d a lio n , F i f t h C e n tu r y ( M a s s o n , I C S 2 , n o . 2 1 7 , I f f .) :

kaspai ewfetasatu basileus kas aptolis ... dowenai...

‘( W h e n t h e M e d e s ... b e s i e g e d t h e c ity o f I d a l i o n , th e k i n g S t a s i k u p r o s a n d
[ikas] t h e c i t y i n v i t e d O n a s i k u p r o s ... t o tr e a t t h e m e n w o u n d e d i n b a t t l e w i t h ­
o u t fe e ). A n d (kaspai) th e k i n g a n d th e c ity p r o m i s e d ... to g iv e . . . ’

(2 0 ) i b id ., 1 0 ff.:

Έ ke sis Onasilon Έ tos kasigriStos ... ex töi khörui töide ex oriae, ide pai o ex
oruxS peisei Onasilöi kas tois kasignStois ... Kas Onasilm oiwöi aneu tö(n)
kasignSton tön ailon ewretasatu basileus kas a ptolis ... dowenai...

24 Note that if the proposed derivation of δέ from δή is correct (cf. e.g. Leumann 1949) this makes
it likely that the original meaning o f the particle was not adversative.
66 ANNA MORPURGO DAVIES

‘i f a n y o n e r e m o v e s O n a s i l o s a n d h i s b r o th e r s ... f r o m t h i s p l a c e ... ( ide pai )

h e w h o r e m o v e s th e m w i l l p a y ... A n d ( kas) t o O n a s i l o s a l o n e w i t h o u t th e
o t h e r b r o th e r s th e k i n g a n d th e c i t y p r o m i s e d to g iv e . . . ’

(2 1 ) ib id ., 2 6 f f .:

ide ta(n) dalton ta(n)de .... basileus kas aptolis katethijan i(n) ta(n) thion tan
Athanan ...
‘(ide) th e k i n g a n d th e c ity p u t th is t a b l e t ... i n th e t e m p l e ( o f ) A t h e n a . . . ’

I n a l l G r e e k te x ts , in c l u d i n g A r c a d ia n , t h e s e c o n d s e n te n c e o f ( 2 0 ) w o u l d b e l i k e l y to
h a v e a δ έ r a t h e r th a n t h e i n t r o d u c t o r y κ ά ς o f th e C y p r i o t te x t. I n a ll l i k e l i h o o d a δ έ
w o u l d a ls o a p p e a r i n ( 1 9 ) a n d i n ( 2 1 ) i n m o s t d i a le c ts , t h o u g h e a r l y A r c a d i a n w o u l d
p r o b a b l y h a v e a n a s y n d e t o n . W e d o n o t h a v e e n o u g h e v i d e n c e t o b e c e r ta in a b o u t th e
r o l e o f ί δ έ i n C y p r io t; i t s a p p e a r a n c e a t t h e s t a r t o f th e a p o d o s is i n ( 2 0 ) is s t r ik in g . I n
a n y c a s e i t is c l e a r l y n o t a d v e r s a tiv e . R u i j g h ( 1 9 6 7 : 5 7 ) , G u s m a n i ( 1 9 6 7 : 1 9 - 2 2 ) a n d
P e r p i l l o u ( 1 9 7 8 : 1 7 9 f f .) w o u l d a ll a g r e e , e v e n i f i n o t h e r r e s p e c t s t h e y r e a c h d if f e r e n t
c o n c lu s io n s .

F r o m th is e v i d e n c e i t i s n o t t o o d i f f i c u l t t o s u r m i s e t h a t th e c o n f l i c t o f κ ά ς / κ α ί a n d
δ έ i n th e p o s t - M y c e n a e a n p e r i o d w a s r e s o l v e d b y C y p r i o t w i t h th e d i s a p p e a r a n c e o f
δ έ ( e x c e p t f o r th e c o m p o u n d e d f o r m s ) a n d w i t h t h e ta k e o v e r b y κ ά ς o f p a r t s o f its
r o l e — l e a v i n g a s i d e ί δ έ b e c a u s e o f o u r ig n o r a n c e . 25 I n A r c a d i a n t h e r e v e r s e h a p ­
p e n e d . A t a lo c a l le v e l (w ith in th e m a in s e c tio n s o f th e te x t) κ ά ς / κ α ί a n d δ έ w e r e
c o n t r a s t e d w h i l e a s y n d e t o n h a d a r o l e t o p l a y to o . T h e e n d r e s u l t w a s t h a t κ ά ς / κ α ί
w a s k e p t w i t h a n a d d i t i v e v a l u e a n d δ έ w i t h a c o n t r a s t i v e v a l u e , w h i l e th e n e u t r a l
t r a n s itiv e r o l e w a s l e f t t o lo c a l a s y n d e to n .

6. The development of δέ in fourth century Arcadian

C l e a r l y w e h a v e o f f e r e d a s p e c u la tiv e a c c o u n t b u t n o t a n i m p l a u s i b l e o n e . T h e n e x t
q u e s t i o n n o w c o n c e r n s t h e f u r th e r d e v e l o p m e n t s o f f o u r th c e n t u r y A r c a d ia n . H o w d o
w e e x p l a i n t h e n e w r o l e o f δ έ a s a q u a s i - s e n t e n c e m a r k e r a n d t h e lo s s o f i t s a d v e r s a -
t i v e / c o n t r a s t i v e v a l u e ? 26 O n e p o s s i b l e a s s u m p t i o n w o u l d o f c o u r s e b e t h a t i n a n
i n c r e a s i n g l y l i t e r a t e s o c i e t y a h i g h e r l e v e l o f s o p h i s t i c a t i o n le a d s to a d i f f e r e n t s e g -

25 It is not uninteresting that, if we accept Warren Cowgill's (1964) interpretation o f the two
crucial passages o f the Idalion inscription, two strongly adversative sentences start with eduwan oin
nu 'they gave ...' and edPk' oin nu 'he gave ...’ respectively. Here the role o f highlighting the
contrapposition is left to word order (the initial position of the verb) and to the oin/ow particle, to
which Denniston (1954: 416) attributes the initial role of stressing the idea of actuality or
essentiality.
26 The phrase Km’ ε ΐ δέ riv a τρόπον which we find in Arcadian inscriptions (IG V 2 6, 1,17, 27)
obviously includes a δέ which is not a connective. Dubois (1986: i, 233) compares the Boeotian
καθ' δν δεί u v a τρόπον and implies that δέ stands for δή. Whatever the explanation of the phrase
its existence proves that δέ was no longer felt as having an adversative value.
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 67

m e n ta tio n o f d is c o u r s e s o th a t th e d e v e lo p m e n t u n its m a r k e d b y a s y n d e to n a re
in c r e a s i n g l y c o m i n g to b e i d e n t i f i e d w i t h t h e w h o l e te x t. T h e r o l e o f δ έ t h e n s im p ly
b e c o m e s t h a t o f g u a r a n te e in g t h e b a s i c s e g m e n ta tio n o f t h e t e x t ( w h ic h h a p p e n s t o b e
i n t o s e n t e n c e s ) a n d th e li n k s b e t w e e n t h e v a r io u s c o m p o n e n t s . T h is is c e r t a i n l y c o r ­
r e c t b u t m a y n o t b e w h o le tr u th . I t i s n o t c e r ta in i n f a c t t h a t t h e d e v e l o p m e n t c a n b e
w h o lly e x p l a i n e d w ith in th e s y s te m . I f w e l o o k a g a in a t ( 5 ) a b o v e , th e p r o x e n y d e c r e e
f r o m w h i c h w e s ta r te d , w e m u s t o b s e r v e t h a t th e l a n g u a g e a n d a r r a n g e m e n t o f th e s e
d e c r e e s , b e y o n d t h e p h o n e t i c a n d l e x i c a l a p p e a r a n c e o f th e l o c a l d i a l e c t , a r e r e m a r k ­
a b ly s i m i l a r i n d i f f e r e n t r e g io n s . T h i s c a n n o t b e d u e e n t i r e l y to c h a n c e ; th e te c h n i c a l
f o r m u l a e w h i c h d e f in e th e h o n o u r s a n d p r iv ile g e s g r a n t e d t o a f a v o u r ite f o r e i g n e r a re
o b v i o u s l y b o r r o w e d f r o m c o m m o n s o u r c e s . T h u s w e d o n o t f i n d i t s u r p r i s i n g i f in
p la c e s as d iv e r s e as C o rc y ra , M e g a r a a n d O r o p o s , w ith d ia le c ts w h ic h r a n g e fro m
C o r in th ia n t o W e s t I o n ic , w e f i n d t h e s e c o n d c la u s e o f a p r o x e n y d e c r e e s t a r t i n g w ith
t h e s a m e δ έ w h i c h w e f o u n d i n ( 5 ) . 27 T e r m i n o l o g y a n d p h r a s e o l o g y c a n b e b o r ­
r o w e d b u t c o n s t r u c t i o n s c a n b e b o r r o w e d to o . I f s o , t h o u g h w e c a n n o t p r o v e i t , i t is
c o n c e i v a b l e t h a t t h e s h i f t i n th e u s e o f δ έ w h i c h w e f i n d i n f o u r t h c e n t u r y A r c a d i a n
i n s c r ip tio n s w a s , i n p a r t a t le a s t, d u e t o e x t e r n a l in f lu e n c e . T h e o b je c tio n th a t p a r tic le s
o r u s e o f p a r tic le s a re n o t s u b je c t to e x te r n a l in f lu e n c e is e a s ily c o u n te r a c te d fo r
A r c a d ia n ; i n t h e sunoikia o f M a n t i n e a w i t h th e H e l i s w a s i o i (n o te 1 8 ), i.e i n t h e m id d le
o f a t e x t ( l i n e 2 1 ) w h ic h i s w r i t t e n i n g o o d d ia le c t, w e f i n d a v e r y n o n - A r c a d i a n "A v
δ έ , i.e . ε ά ν δ έ , f o r w h a t i n A r c a d i a n w o u l d b e ε ί. δ ' α ν .

7, μ έ ν ... δ έ

T h e r e i s p e r h a p s m o r e . S o f a r w e h a v e l e f t a s id e μ έ ν a n d t h e μ έ ν ... δ έ c o n s t r u c t i o n .
T h is is l a r g e l y b e c a u s e μ έ ν is r a r e i n A r c a d ia n . I f i t o c c u r s , i t is o n l y i n s t a n d a r d t e x t­
b o o k c o n s t r u c t i o n s o f th e μ έ ν ... δ έ t y p e w h i c h a p p e a r s o m e w h a t s tilte d ;

( 2 2 ) R e g u la tio n s f o r th e te m p le o f A th e n a A le a ( n o te 1 4 ), 1 .1 5 ;

... τ ο μ έ ν μ έ ζ ο ν π ρ ό β α τ ο ν δ α ρ χ μ ά ν ό φ λ έ ν , to δ έ μ έ ΐο ν Ιν φ ο ρ β ίεν
‘o n t h e o n e h a n d ( μ έ ν ) f o r a b i g g e r p i e c e o f c a t t l e h e w i l l o w e a d r a k h m a , o n
th e o th e r h a n d (δ έ ) f o r a s m a lle r o n e th e re w ill b e iixphorbismos'

(2 3 ) i b id ., 1 . 2 2

to μ έ ν δ μ ισ υ r a t θ ε ο ί, tò δ έ ε μ ι σ υ τ ο ι ς Ι ι ι ε ρ ο μ ν ά μ ο ν σ ι ...

27 The examples are chosen al random: Corcyra IG IX 1 682, Schwyzer DGE 136 (fourth century):
πρόξενον π ο ιε ί ά άλία ...· δίδωτι δέ κα ί γας κα ι οικ ία ς έμπασιν ..., Megara IG VII 8,
Schwyzer DGE 155 (early third century): ... πρόξενον αύτόν εΐμεγ καί έκγόνους αύτοΰ ...·
έΐμεν δέ αιηώ κα ί οΙκίας έμπασιν, Oropos IG VII 4250, Schwyzer DGE 812 (fourth century):...
Άμύνταν Ά ντιόχου Μακεδόνα πρόξενον εΐν ..." ατέλειαν δέ ειν κα ί άσυλιαν ...
 ANNA MORPURGO DAVIES

o n t h e o n e h a n d ( μ έ ν ) h a l f t o t h e g o d d e s s , o n t h e o t h e r h a n d ( δ έ ) h a l f to
t h e t e m p le a d m i n i s t r a t o r s ... ’

( 2 4 ) R u le s a b o u t b u i l d i n g c o n tr a c ts ( n o te 1 5 ) , 1. 4 9 :

... τ ο ν μ έ ν έ ρ γ ά τ α ν έ σ δ έ λ λ ο ν τ ε ς ... , τ ο ν δ έ έ ρ γ ώ ν α ν ζ α μ ν ό ν τ ε ς .. .
‘( l e t t h e m h a v e p o w e r ) o n t h e o n e h a n d ( μ έ ν ) t o s a c k t h e w o r k m a n , o n t h e
o th e r h a n d ( δ έ ) to fin e th e c o n tra c to r ...’

( 2 5 ) S y n o ik ia b e t w e e n O r c h o m e n o s a n d E u a i m o n ( n o te 1 7 ), 1. 7 3 a n d 1. 9 2 :

... κ ε ύ ο ρ κ έ ν τ ι μ έ ν τ ά γ α θ ά , [ έ ] π ι ο ρ κ έ ν τ ι δ έ έ ξ ο [ λ έ ] σ θ α ι α ύ τ ό γ κ α ί γ έ ν ο ς
‘... o n th e o n e h a n d ( μ έ ν ) f o r w h o e v e r r e s p e c ts t h e o a t h a l l g o o d t h in g s , o n th e
o t h e r h a n d ( δ έ ) f o r w h o e v e r b r e a k s t h e o a t h l e t h i m d ie a n d h is d e s c e n d a n t s '

( 2 6 ) D e c r e e f o r th e r e t u r n o f e x ile s ( n o t e 1 6 ), 1. 4 1 ff.:

... ε ί μ έ ν α ν φ α ί ν η τ ο ι ... ε ί δ ’ α ν μ ή φ α ί ν η τ ο ι ...

‘( w h o e v e r h o l d s t h e p r o p e r t y ) o n t h e o n e h a n d ( μ έ ν ) i f h e l o o k s a s ( h a v i n g
p a i d th e d e b t t o th e g o d d e s s ... l e t h i m g i v e b a c k t o t h e r e t u r n i n g e x i l e h a l f o f
th e p r o p e r t y ) , i f o n t h e o t h e r h a n d ( δ έ ) h e d o e s n o t lo o k a s ( h a v in g g iv e n b a c k
t o th e g o d d e s s ) , . . ’28

I n o t h e r t e x ts a n d o t h e r d ia l e c t s μ έ ν is f a r m o r e f r e q u e n t a n d s e e m s t o b e u s e d w i t h
m o r e f re e d o m . S h o u ld w e a s s u m e th a t w e a r e d e a lin g w ith a n A r c a d ia n in h e r ite d
u s a g e ? A b o u t f i f t y y e a r s a g o M a n u L e u m a n n ( 1 9 4 9 ) a r g u e d w i t h g o o d e v id e n c e t h a t
μ έ ν w a s s im p ly a s h o r t e n e d f o r m o f μ ή ν , w h i c h i n its tu r n w a s th e I o n i c f o r m o f μ ά ν .
D e n n i s t o n d i e d i n t h e s a m e y e a r i n w h i c h L e u m a n n ’s a r t i c l e a p p e a r e d a n d c o n s e ­
q u e n tly w e d o n o t k n o w h o w h e w o u ld h a v e re a c te d ; h e d id p o in t o u t ([1 9 3 4 ] 1 9 5 4 :
3 2 8 ) t h a t ‘t h e p a r a l l e l i s m i n t h e u s e s o f μ ά ν ( μ ή ν ) a n d μ έ ν i s o n t h e w h o l e r e m a r k ­
a b l y c l o s e ’ b u t a l s o o b s e r v e d th a t, w h ile , l e a v i n g E p i c o n th e o n e s id e , ‘μ ά ν , μ ή ν , μ έ ν
a r e c o n f i n e d r e s p e c t i v e l y t o D o r i c , A t t i c a n d I o n i c 1, ‘p r e p a r a t o r y μ έ ν is c o m m o n to
a l l t h r e e d i a l e c t s ’. H o w e v e r , i f L e u m a n n i s r i g h t , i t s h o u l d f o l l o w t h a t μ έ ν m u s t b e
A ttic o r I o n ic s in c e it p re s u p p o s e s a c h a n g e α > η . A r c a d ia n u s a g e d o e s n o t c o n tr a ­
d i c t th i s v ie w : a s p o i n t e d o u t a b o v e t h e o n l y c o n s t r u c t i o n s a t t e s t e d a r e t h o s e w i t h
p r e p a r a t o r y μ έ ν a n d t h e s e s h o w a l a c k o f f l e x i b i l i t y w h i c h is s tr i k i n g ; i f th e y w e r e
d u e t o th e im i t a t i o n o f a n o u t s i d e m o d e l t h e t e x t b o o k f e e l i n g w h i c h I m e n t i o n e d e a r ­
l i e r w o u l d b e u n d e r s ta n d a b le .

28 The sequence δ μέν ... δ δέ probably appears also in a very fragmentary text from Mantinea of
the early fifth century (IG V 2 261) which seems to offer the first instance o f μέν.
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 69

8. Some conclusions

I s ta r te d i n 1. w i t h th r e e q u e s tio n s o r o b s e r v a tio n s t o w h i c h w e n o w n e e d to r e tu r n .

M e r e ly f r o m t h e f e w e x a m p le s a d d u c e d i t is c l e a r th a t t h e e p i g r a p h ic a l te x ts w h i c h w e
h a v e b e e n lo o k i n g a t d o n o t o f f e r th e s a m e e x te n s iv e u s e o f p a r tic le s w h i c h w e f i n d i n
H o m e r o r A t t i c p r o s e . O f c o u r s e t h e v e r y l i m i t e d u s e o f p a r t i c l e s i n A r c a d ia n i n s c r i p ­
tio n s d o e s n o t p r o v e t h a t th e s p o k e n l a n g u a g e w a s e q u a l l y d e p l e t e d o f p a r t i c l e s . [E d .
n o te : s e e f o r t h i s i s s u e a ls o D u h o u x ’ a r tic le i n th is v o l u m e ] . B u t m o r e w o r k n e e d s to
b e d o n e t o a s c e r t a i n h o w f a r th is a p p a r e n t p o v e r ty i s c h a r a c t e r i s t i c o f s p e c if ic d ia le c ts
o r s p e c i f i c p e r i o d s o r o f e p i g r a p h i c a l l a n g u a g e i n g e n e r a l . O n e th i n g , h o w e v e r , is
c e r ta in : w e c a n n o t a s s u m e apriori t h a t a ll G r e e k d ia le c ts s h a r e d th e s a m e p a rtic le s
a n d m a d e t h e s a m e e x te n s iv e u s e o f th e m .

T h e h is to ry o f A rc a d ia n o n th e o n e h a n d a n d th a t o f A rc a d o -C y p rio t a n d M y c e n a e a n
o n t h e o t h e r m a k e c l e a r t h a t d e v e l o p m e n t s i n t h e u s e o f p a r t i c l e s , i.e . c h a n g e s i n th e
p a r t i c l e in v e n t o r y a s w e l l a s c h a n g e s i n t h e w a y i n w h i c h t h e s a m e p a r t i c l e s a r e u s e d
a n d in t h e f u n c t i o n t h a t th e y f u lf il, a r e c h a r a c te r is tic n o t o n l y o f lite r a r y la n g u a g e s b u t
a l s o o f th e e p ig r a p h i c a l l y a t t e s t e d d i a l e c t s . T h e le s s o n t o b e l e a r n e d i s t h a t w e s h o u ld
b e c a re fu l in a ttr ib u tin g to th e o n e o r th e o th e r d ia le c t a s p e c if ic p a rtic le o r p a r tic le
u s e ; i t i s c o n c e i v a b l e t h a t i n e a c h i n s t a n c e w e a r e d e a l i n g w i t h f e a tu r e s w h i c h b e l o n g
t o a s p e c if ic p h a s e r a t h e r th a n to th e w h o l e h i s t o r y o f t h e d ia le c t. T h e i m p o r t a n t p o i n t
is t h a t e v e n i n o u r s m a ll w o r l d w e h a v e e v id e n c e n o t o n l y f o r th e lo s s o f o l d p a r t i c l e s
b u t a ls o f o r t h e a p p e a r a n c e o f n e w p a r tic le s .
T h e w a y i n w h i c h th e s e c h a n g e s o c c u r p a r t l y c o n t r a d i c t s t h e r e c e i v e d o p i n i o n ,
b u t o n l y p a r t l y . T h e p r o c e s s e s w h i c h l e d t o th e d i s a p p e a r a n c e o f δ έ i n C y p r i o t a n d to
th e e m p h a s i z i n g o f i t s a d v e r s a t i v e v a l u e in e a r l y A r c a d i a n a r e l i k e l y t o b e in t e r n a l l y
l e d ; i t w o u l d b e d i f f i c u l t t o s e e h o w th e y c o u l d h a v e b e e n d e t e r m i n e d b y e x t e r n a l
i n f l u e n c e s . O n t h e o t h e r h a n d , w e s h o u l d n o t n e c e s s a r i l y t h i n k i n te r m s o f p u r e l y
s y s t e m i c d e v e l o p m e n t s ; i t m a y w e l l b e th a t, a s B a k k e r ( 1 9 9 3 ) h a s s u g g e s t e d , th e
a d v e r s a tiv e v a l u e o f δ έ is c o n n e c te d w i t h th e d e v e lo p m e n t o f a w r itte n c o n t e n t- b ia s s e d
t r a d itio n w h i c h r e p l a c e s a n o l d o r a l tr a d itio n . A t t h e s a m e t i m e i t w o u l d b e f o o l i s h to
e x c l u d e t h e i m p o r t a n c e o f i n t e r d i a l e c t a l c o n t a c t s f o r t h e d e v e l o p m e n t o f th e p a r t i c l e
s y s te m . T h e s t a n d a r d e x a m p le h a s b e e n a v a ila b le f o r a l o n g tim e . W e a s s u m e t h a t th e
s p re a d o f κ α ί a n d th e q u a s i-d is a p p e a ra n c e o f τ ε a re p o s t-M y c e n a e a n ; it fo llo w s th a t
e i t h e r th e y h a p p e n e d i n d e p e n d e n t l y i n a l l G r e e k d i a l e c t s — w h ic h i s h a r d l y p l a u s i ­
b le — o r t h a t w e m u s t a llo w f o r in f l u e n c e s a n d c o u n t e r i n f l u e n c e s o f v a r i o u s ty p e s . In
th e te x ts t h a t w e h a v e s tu d ie d t h e r e p l a c e m e n t o f M a n t i n e a n κ ά ς w ith κ α ί g i v e s c o n ­
c r e t e e v i d e n c e f o r a s i m i l a r p r o c e s s , i . e . t h e i n t r o d u c t i o n o f a n e w f o r m o f a w e ll
k n o w n p a r t i c l e t h a n k s to e x t e r n a l in f l u e n c e . B u t d o w e h a v e e v i d e n c e f o r th e w h o l e ­
s a le b o r r o w i n g n o t o f a n e w f o r m o f p a r t i c l e b u t o f a n e w p a r t i c l e a n d p e r h a p s o f n e w
c o n s t r u c t i o n s ? A b o v e I h a v e a r g u e d t h a t b o t h th e m o s t p l a u s i b l e e ty m o l o g y a n d th e
s o m e w h a t s tilte d , te x tb o o k - lik e , u s e o f μ ε ν i n A r c a d ia n s p e a k f o r a b o r r o w e d p a r tic le .
I f t h i s i s c o r r e c t , i t a l s o f o l l o w s , a s L e u m a n n b e l i e v e d , t h a t a l l i n s t a n c e s o f ‘D o r i c ’
70 ANNA MORPURGO DAVIES

a n d ‘A e o l i c ’ e p i g r a p h i c a l μ έ ν a r e d u e t o b o r r o w i n g s a n d t h a t th e s e b o r r o w i n g s a r e
r e a s o n a b l y e a r ly : j u s t t o g iv e a n e x a m p le , μ έ ν ... δ έ a r e f o u n d i n s i x t h c e n tu r y C r e t a n
in s c r ip tio n s a n d a r e o m n ip r e s e n t in th e G o r ty n c o d e . S im ila r ly I h a v e a ls o m a d e a
c a s e f o r t h e th e p o s s i b i l i t y t h a t t h e p u r e l y c o n n e c t i v e / t r a n s i t i v e u s e o f δ έ w h i c h
a p p e a r s i n th e A r c a d i a n in s c r ip tio n s o f m i d a n d l a t e f o u r th c e n t u r y m a y b e in f l u e n c e d
f r o m o u ts id e A r c a d ia .

D o w e t h e n r e j e c t th e v i e w t h a t p a r tic le s b e l o n g to a c l o s e d c l a s s a n d a r e n o t p r o n e t o
b o r r o w i n g o r d o w e t a k e r e f u g e i n th e o b s e r v a t i o n t h a t w e a r e d e a l i n g w i t h d ia l e c t s
a n d n o t w i t h s e p a r a t e la n g u a g e s , w h i c h c o u l d e x p l a i n w h y t h e d e v e l o p m e n t s w h i c h
w e h a v e b e e n d e s c r ib in g m a y b e e x te r n a lly d e te rm in e d ? T h e g e n e r a l p o in t re m a in s
u n c h a l l e n g e d . T h e s ta tu s o f th e G r e e k d ia l e c t s a s d ia le c ts o r l a n g u a g e s , i f th e d is tin c ­
t i o n i s m e a n in g f u l, i s o f c o u r s e i n d is p u te , b u t th e r e is little d o u b t t h a t s tr u c tu r a l s im i­
la r ity a n d fre q u e n t c o n ta c ts m u s t h a v e a llo w e d a ra n g e o f m u tu a l in flu e n c e s w h ic h
w o u ld n o t h a v e b e e n p o s s ib le in d iffe re n t s o c io lin g u is tic s itu a tio n s . H o w e v e r, w e
o u g h t a t le a s t to n o t i c e t h a t th e g e n e r a l p r i n c i p l e i t s e l f h a s b e e n p o w e r f u l l y c h a lle n g e d
b y e v i d e n c e m o r e c o m p e l l i n g th a n t h a t d i s c u s s e d a b o v e . I n c o n s i d e r i n g a s e r ie s o f
c l a i m s a b o u t th e p la u s i b i l i t y o r o th e r w is e o f g r a m m a tic a l b o r r o w i n g C a m p b e ll ( 1 9 9 3 )
c o n c l u d e s (p . 1 0 3 f .) t h a t ‘n o n e o f th e m h o l d s t r u e i n a n y a b s o l u t e s e n s e . ’ I n p a r t i c u ­
l a r h e r e je c ts th e v i e w t h a t ‘p r e p o s itio n s , c o n ju n c tio n s , a n d p a r t i c l e s , to t h e e x te n t th a t
t h e y a r e u s e d g r a m m a t i c a l l y , c a n b e b o r r o w e d f r o m o n e l a n g u a g e to a n o t h e r o n l y
w i t h g r e a t d if f ic u lty , ’ p o in t i n g t o th e c o u n te r v id e n c e p r o v i d e d b y ‘t h e ty p ic a l b o r r o w ­
i n g o f c o n ju n c tio n s a n d o t h e r d is c o u r s e p a r t i c l e s f r o m S p a n is h i n t o a la r g e v a r i e t y o f
I n d i a n la n g u a g e s o f L a t i n A m e r i c a ’ ( i b i d ., 1 0 0 ) . I n a n e a r l i e r a r t i c l e B r o d y ( 1 9 8 7 )
h a d a n a l y z e d t h e e x t e n s i v e b o r r o w i n g f r o m S p a n i s h in t o t h e M a y a n la n g u a g e s o f
p a r t i c l e s u s e d a s d i s c o u r s e m a r k e r s a n d p o i n t e d to v a r io u s p r o c e s s e s w h i c h l e d e ith e r
t o t h e j o i n t u s e o f i s o f u n c t i o n a l S p a n i s h a n d M a y a n p a r t i c l e s o r t o th e e x c l u s i v e u s e
o f S p a n is h p a rtic le s lik e pero ‘b u t ’ , corno ‘l i k e ’, pues ‘w e l l ’, y ‘a n d ’, porque
‘b e c a u s e ’, entonces ‘t h e n ’, a l l o f w h i c h c a n f u n c t i o n a s d i s c o u r s e m a r k e r s i n M a y a n .
H e r e a r e w e d e a lin g w i t h u n r e la te d a n d s tr u c tu r a lly d if f e r e n t l a n g u a g e s , o n e o f w h i c h
h o w e v e r is s o c i o lin g u is tic a lly d o m in a n t. O n th e o t h e r h a n d , a t a m o r e a n e c d o ta l l e v e l,
i t is n o t t o o d i f f i c u l t t o f i n d s i m i l a r e x a m p l e s n e a r e r h o m e : i t is g e n e r a l l y a s s u m e d
t h a t th e M o d e r n G r e e k μ ά ‘b u t ’, w h i c h r e p l a c e s th e o ld ά λ λ α a n d t o a c e r ta in e x t e n t
δ έ , 29 i s a b o r r o w i n g f r o m I t a l i a n ma ‘b u t ’, i n i t s tu r n a r e p l a c e m e n t f o r L a t i n sed
d e riv e d fro m L a tin magis ‘m o r e , r a t h e r ’.

T h e f i n a l c o n c lu s io n is t h a t w e s h o u l d n o t h e s i t a t e to a s s u m e t h a t i n th e c l a s s o f p a r ­
t i c l e s in te r d ia le c ta l b o r r o w in g s a r e n o t o n l y p o s s i b l e b u t lik e ly . I f s o , h o w e v e r , m u c h
r e m a i n s to b e d o n e t o e s ta b lis h n o t o n l y w h a t p a r t i c l e f o r m s w e r e b o r r o w e d f r o m o n e
d i a l e c t in t o a n o t h e r b u t a l s o h o w p a r t i c l e u s a g e w a s i n a n y s p e c i f i c c a s e d e t e r m i n e d
b y e x t e r n a l i n f l u e n c e s . I f i t i s in d e e d t h e c a s e , a s f o r M a y a n a n d S p a n i s h , t h a t th e
d o m i n a n t la n g u a g e i s m o r e lik e ly t o i n f l u e n c e t h e u s a g e o f th e le s s i m p o r t a n t la n -

29 Not to be confused with Thessalian μά, which in that dialect fulfils the functions of δέ.
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 71

g u a g e , s tu d ie s o f th is k i n d m a y p r o v i d e u s w i t h im p o r t a n t in s ig h ts a b o u t t h e s o c io lin -
g u is tic s ta tu s o f th e G r e e k d ia le c ts .

References

A n d r e o u , I.
1986 ‘Τ α ε π ι γ ρ ά μ μ α τ α τ ο ύ π ο λ υ α ν δ ρ ί ο υ τ η ς Α μ β ρ α κ ί α ς ’. Α ρ χ . Δ ε λ τ . 4 1 α \
Μ ε λ έ τ ε ς 4 2 5 -4 6 , Π ίν . 9 7 -1 0 0
B a k k e r , E .J .
1993 ‘B o u n d a r i e s , T o p i c s a n d th e S tr u c t u r e o f D i s c o u r s e . A n I n v e s t i g a t i o n o f th e
A n c i e n t G r e e k P a r t i c l e d e ’. Studies in Language 1 7 ,2 7 5 - 3 1 1
B o u s q u e t, J.
1992 ‘D e u x é p i g r a m m e s g r e c q u e s ( D e lp h e s , A m b r a c i e ) ’. BCH 1 1 6 , 5 8 5 - 6 0 6 , a t
5 9 6 -6 0 6
B r o d y , J.
1987 ‘P a r t i c l e s B o r r o w e d f r o m S p a n i s h a s D i s c o u r s e M a r k e r s i n M a y a n L a n ­
g u a g e s ’. Anthropological Linguistics, 2 9 , 5 0 7 -2 1
C a m p b e ll, L .
1993 O n P r o p o s e d U m v e r s a l s o f G r a m m a t i c a l B o r r o w i n g ’. In : H . A e r t s e n a n d
Historical Linguistics 1989. Papers from the 9th Inter­
R .J . J e f f e r s ( e d s ) ,
national Conference on Historical Linguistics. A m s t e r d a m , 9 1 - 1 0 9
C a s s io , A . C .
1994 ‘I d i s t i c i d e l Polyandrion d i A m b r a c i a e l ' “ i o a n o n i m o ” n e l l 'e p i g r a m m a
g r e c o ’ . SMEA 3 3 , 1 0 1 - 1 1 7
C o w g ill, W .
1964 ‘T h e S u p p o s e d C y p r i o t O p t a t i v e s duwànoi a n d dökoi. W ith N o te s o n th e
G r e e k I n f i n i t i v e F o r m a t i o n s ’. L a n g u a g e 4 0 , 3 4 4 - 6 5
D e n n i s t o n , J .D .
1954 The Greek Particles. 2 n d e d . O x fo rd (1 s t ed. 193 4 )
D G E c f. S c h w y z e r
D u b o is , L .
1986 Recherches sur le dialecte arcadien. 3 v o ls ., L o u v a in - la - N e u v e
D u n k e l, G .
1990 ‘J . W a c k e m a g e l u n d d i e i d g . P a r t i k e l n * j a , *ke, *kem u n d *an’. I n : E .
E i c h n e r u n d H . R ix ( H r s g .) , Sprachwissenschaft und Philologie. Jacob
Wackernagel und die Indogermanistik heute. W i e s b a d e n , 1 0 0 - 1 3 0
E g e tm e y e r , M .
1992 Wörterbuch zu den Inschriften im kyprischen Syllabar. B e r lin
G u s m a n i, R .
1967 ‘Z w e i k y p r i s c h e K o n j u n k t i o n e n ’. Glotta 4 5 , 1 9 - 2 5
H a n s e n , P .A .
72 ANNA MORPURGO DAVIES

1 9 8 3 -8 9 Carmina epigraphica graeca. 2 v o ls . , B e r lin - N e w Y o r k

H augen, E.
1950 ‘T h e A n a ly s is o f L i n g u is tic B o r r o w i n g ’. Language 2 6 ,2 1 0 - 3 1
H in tz e , A .
1993 A Lexicon to the Cyprian Syllabic Inscriptions. H a m b u r g
H o d o t, R .
1990 O r g a n is a tio n te x tu e lle e t p a r tic u le s m o d a le s d a n s d e s te x te s le g is la tifs
a r c a d ie n s ( V e - I V e S . a . C . ) \ Verbum 13, 2 7 -3 7
H u m b e r t , J.
1954 Syntaxe grecque. P a r is
I C S c f. M a s s o n 1 9 8 3
J a m ie s o n , J.
1814 Hermes Scythicus or, The Radical Affinities o f the Greek and Latin Lan­
guages to the Gothic. E d in b u r g h
K ro o n , C.
1995 Discourse Particles inLatin. A m s te r d a m
K ü h n e r , R . - B . G e r th
1904 Ausftirhliche Grammatik der griechischen Sprache, Zweiter Teil: Satzlehre.
2 v o ls . H a n n o v e r a n d L e ip z ig
L eum ann, M .
1949 ‘μ έ ν u n d μ ή ν , δ έ u n d δ ή ’. Museum Helveticum 6 , 8 5 -8 9 (R e p rin te d in
Kleine Schriften. Z ü r ic h u n d S tu ttg a rt 1 9 5 9 , 2 2 9 -3 4 )
L e v in s o h n , S.
1987 Textual Connections in Acts. A d a n t a , G e o r g ia
L ü tte l, V .
1981 Κ ύς und Kai. Dialektale und chronologische Probleme im Zusammenhang
mit Dissimilation undApokope. E r g s h . z u r K Z N r 2 9 , G ö t t i n g e n
M asso n , O.
1983 Les inscriptions chypriotes syllabiques [IC S ]. R é im p r e s s io n a u g m e n té e .
P a r is
M a s s o n O . e tT .B . M itf o r d
1986 Les inscriptions syllabiques de Kouklia-Paphos. K o n s t a n z
M o i p u r g o D a v ie s , A .
1992 ‘M y c e n a e a n , A r c a d i a n , C y p r i a n a n d s o m e Q u e s t i o n s o f M e t h o d i n D i a l e c ­
t o l o g y ’. I n : J .P . O li v i e r ( e d .) , Mykenaika. B C P I S u p p i. X X V , 4 1 5 -3 1
N eum ann, G.
1975 ‘Z u d e n H e x a m e t e r n d e r k y p r i s c h e n I n s c h r i f t I C S 2 6 4 ( m i t e i n e m B e i t r a g
v o n K la u s S t i e w e ) ’. Kadmos 1 3 , 1 4 6 - 5 5
1994 Ausgewählte Kleine Schriften. I n n s b r u c k
P e r p i l l o u , J .-L .
1978 ‘D é b u t s d e p h r a s e e n g r e c m e r i d i o n a l a r c h a i'q u e ’. I n : Etrennes de septan-
taine. Travaux de linguistique et de grammaire comparée offerts à Michel
Lejeune. P a r i s , 1 7 9 - 1 8 1
 PARTICLES IN GREEK EPIGRAPHICAL TEXTS 73

R is c h , E .
1969 ‘D i e v e r s c h i e d e n e n P a r t i k e l n δ ε i m G r i e c h i s c h e n ’. I n : Studi linguistici in
onore di Vittore Pisani. V o i . 2 , B r e s c i a , 8 3 1 -4 3 (R e p rin te d in Kleine
Schriften. B e r lin - N e w Y o r k 1 9 8 1 , 4 8 6 - 9 8 )
1988 ‘L e d é v e l o p p e m e n t d u c h y p r i o t e d a n s l e c a d r e d e s d i a l e c t e s g r e c a n c i e n s ’.
In : J. K a r a g e o r g h is a n d O . M a s s o n ( e d s ), The History o f the Greek Lan­
guage in Cyprus. N ic o s ia , 6 7 -7 7
R u i j g h , C .J.
1957 L'élément achéen dans la langue épique. A s s e n
1967 Etudes sur la grammaire et le vocabulaire du grec mycénien. A m s t e r d a m
1971 Autour de τε épique. Éludes sur la syntaxe grecque. A m s te r d a m
1981 R e v i e w o f L ü t t e l 1 9 8 1 . Kratylos 2 6 , 1 1 5 - 2 0
S c h w y z e r, E .
1923 Dialectorum graecarum exempla epigraphicapotiora [D G E ]. L e ip z ig
T e R i e l e , G .- J .- M .- J .
1987 ‘H é l i s s o n e n t r e e n s y m p o l i t i e a v e c M a n t i n é e : u n e n o u v e l l e i n s c r i p t i o n d ’
A r c a d i e ’ . B O T 1 1 1 , 1 6 7 -8 8
T h ü r, G . u n d H . T a e u b e r
1994 Prozessrechtliche Inschriften der griechischen Poleis. W i e n
 Άλλ’έξόλοισθ’ αυτφ κοαξ
Réexamen des emplois de άλλ,ά à la lumière
de Vénonciation
dans Les Grenouilles d’Aristophane

Louis b a s s e t
Université Lumière Lyon 2

1. Les conditions d ’un réexamen

1.1. L ’intérêt d ’une nouvelle approche, pragmatique

L ’étude des conjonctions «adversatives» a été depuis une vingtaine d’années renou­
velée par la prise en compte des conditions de l’énonciation. Le point de départ de
cette nouvelle approche fut une brève remarque de O. Ducrot (1972: 128-9) pro­
posant une interprétation «argumentative» de la conjonction française m a i s . Π a
ensuite lui-même développé cette interprétation, seul ou avec d’autres, en plusieurs
circonstances.1
Des approches semblables, c’est-à-dire de type p r a g m a t i q u e , ont été par la suite
proposées pour des conjonctions adversatives appartenant à des langues diverses.
Citons pour le latin, l’étude de ai, opposé en particulier à s e d , par C. Kroon (1991, et
1995: 333-370), et celles de A. Orlandini sur a t d ’une part, i m m o et a t q u i d’aulTe
part (1994 et 1995). Ces deux auteurs ont en outre la particularité de rattacher leurs
analyses à la F u n c t i o n a l G r a m m a r de S.C. Dik, dans laquelle est en effet proposée
une analyse pragmatique de l ’énoncé (1978: 127-156).
Pour le grec, la liaison μέν ... δέ ... après négation a été étudiée du point de vue
de l ’argumentation par A.-M. Chanet (1990).2 Mais à ma connaissance, la conjonc­
tion αλλά n ’a pas donné lien à une étude de ce genre. Les pages que lui avait con­
sacrées C.J. Ruijgh dans A u t o u r d e τ ε é p i q u e (1971: 125-126, 135-140, 785-794)
sont en effet antérieures au développement de la pragmatique. Elles étaient donc
tributaires de l ’exposé de J.D. Denniston (1954:1-32), tout en y mettant de l ’ordre.3

1 Voir bibliographie jointe.

2 L'emploi de μέν y est décrit comme celui d’un «repoussoir argumentatif» (p. 33).
3 C.J. Ruijgh a regroupé, p. 125, les divers emplois des conjonctions adversatives sous deux
rubriques, l ’emploi oppositi/ (allemand aber) et l’emploi éliminatif (allemand sondern). L ’emploi
éliminalif était considéré, p. 135, à partir des données statistiques d ’Hérodote, comme l’emploi fon­
damental de άλλα. L ’étude proposée ici doit beaucoup à cette mise en ordre, meme si la perspective
76 LOUIS BASSET

Mais certaines remarques anticipaient déjà sur les développements ultérieurs, par
exemple (p. 138):
«Ainsi, quand άλλά introduit un ordre après un discours ou un dialogue, la
particule signale, pour ainsi dire, que le discours ou le dialogue ne suffit pas
mais qu’il faut procéder à Γ action.»
De telles remarques, ainsi que la mention répétée d’emplois «correctifs» de άλλά, ce
qui peut suggérer une interaction verbale, sont une incitation à réexaminer l’ensem­
ble des emplois de άλλά à la lumière des conditions d’énonciation.

1.2. Le choix d ’un dialogue comme texte de référence

Or, si les emplois conjonctifs de άλλά sont liés à l’énonciation en général et aux
intentions argumentatives en particulier, il paraît préférable de les observer dans un
texte ou domine le dialogue. Je me suis donc proposé de les étudier dans une
comédie d’Aristophane, et j ’ai choisi Les Grenouilles, comédie riche en dialogues
agonistiques. L ’exposé de Denniston (1954) me permettait de contrôler l’absence
éventuelle de certains emplois dans ce corpus restreint. Je suivais d’ailleurs dans ce
choix l’exemple même d ’O. Ducrot, avec d ’autres linguistes.
Ses propres conceptions sur la conjonction mais ont en effet été illustrées et
approfondies dans une recherche collective consacrée aux emplois de mais dans deux
scènes d’une comédie de Feydeau. Le résultat de cette recherche fut un article intitulé
«Mais occupe-toi d ’Amélie» publié une première fois en 1976, puis repris dans Les
mots du discours en 1980. Le dessein de cette étude était de tenter une description
unitaire du mais français en rattachant tous ses emplois, sauf toutefois ceux qui cor­
respondent à l ’allemand sondern, au «modèle standard» qui avait été précédemment
défini, et qui était ainsi résumé (1980: 97):
«l’expression P mais Q présuppose que la proposition P peut servir d’argu­
ment pour une certaine conclusion r et que la proposition Q est un argument
qui annule cette conclusion.»
La difficulté de l ’entreprise tenait surtout au fait que, le texte de référence étant
un dialogue comique, de nombreux mais se trouvaient en début de réplique; ce qui
rendait souvent difficile l ’identification de P et de sa conclusion r. Il m’a même
semblé que cette identification avait souvent un caractère acrobatique, qui n’emportait
pas toujours l ’adhésion. Enfin, l ’impossibilité de rattacher au modèle standard les
emplois correspondant à l ’allemand sondern, m’incitait à penser qu’il n’était pas
assez général, pour mais comme pour άλλά.
Aussi ai-je fait mien, pour l ’appliquer au grec άλλά, l’espèce de voeu final qui au
terme de cette étude concluait sur les résultats acquis et les difficultés rencontrées
(Ducrot et al. 1980: 130): que l ’étude soit reprise, pour montrer l ’unité sémantique
de la conjonction étudiée, en privilégiant les emplois du dialogue, en particulier en
début de réplique, et sans Va priori d’une interprétation tirée du discours suivi. Il

pragmatique et le choix d ’un dialogue comme texte de référence ont pour effet d ’inverser la
hiérarchie des emplois.
 'Α λλ' έ ξ ό λ ο ισ β ’ αύτώ κοαξ 77

m’a semblé que, contrairement à l’espoir qui était exprimé, les résultats ne seraient
pas nécessairement concordants.

1.3. Unité sémantique et diversité fonctionnelle

La recherche d ’une description unitaire doit cependant permettre de rendre compte

aussi de la diversité des emplois. Je suis pour cela parti de l’idée, maintenant assez
commune, qu’en la matière l’unité est d ’ordre sémantique et la diversité d ’ordre
fonctionnel. Une signification unique, qu’il faut définir, est à la base de tous les
emplois. Les différences entre ceux-ci viennent des différents objets linguistiques que
peut coordonner la conjonction άλλα. Il ne s’agit donc pas de différences de sens,
comme le suggèrent les distinctions d ’emplois éliminatifs, adversaiifs, progressifsf
H s’agit en fait de différences de portée syntaxique ou pragmatique.
C ’est de cette façon qu’en d’autres temps j ’ai essayé de rendre compte des dif­
férents emplois des particules modales et de certains adverbes comme τάχα, traduit
«bientôt» ou «peut-être» (Basset 1988: 33-34). J’avais par exemple expliqué ces
deux emplois de τάχα par la même valeur de proximité, mais tantôt c’est une
proximité portant sur l ’événement (le contenu de l ’énoncé), tantôt une proximité
portant sur l ’assertion (l’énoncé lui-même, c’est-à-dire son énonciation).
C’est de la même façon que l ’on distingue en français des emplois différents de
certains adverbes, comme sincèrement, adverbe d’énonciation ou de contenu. C ’est
donc sur ce modèle, qui peut cependant être affiné, qu’on envisage ici de distinguer
les emplois de άλλά.

2. Exclusion des emplois non conjonctifs

2.1. Les emplois adverbiaux

Mais il importe cependant de ne retenir que les emplois conjonctifs. Il faut donc
éliminer d ’éventuels emplois adverbiaux de άλλά, ceux qui sont les plus proches du
sens étymologique «autrement». Outre l ’analyse sémantique et le critère de la posi­
tion (initiale pour άλλά conjonction), on peut utiliser ici un critère souvent proposé:
si deux particules relient les mêmes objets linguistiques, l ’une au moins n ’est pas
une conjonction. On a par exemple écarté «donc» de la liste des conjonctions
françaises, puisque «donc», dans tous ses emplois, peut se combiner avec une con­
jonction reconnue comme telle («or donc», «et donc»; alors que *«or car», par
exemple, est impossible (cf. en particulier Dik 1968: 34).
De même, dans la collocation δ' άλλά, présente chez Aristophane, mais pas dans
les Grenouilles, άλλά est une particule adverbiale (Denniston p. 10). En revanche,
dans le tour épique άλλά τε, c’est τε qui est un adverbe, ainsi que l ’a montré C.J.

V oir D enniston p. 1 ( eliminative

), p. 5 ( balancing ), p. 20-21 ( inceptiveadversative,progressive
, ).
 LOUIS BASSET

Ruijgh (1971: 785-794). On trouve aussi ailleurs chez Aristophane la séquence άλλ'
où (...) γάρ dans laquelle c’est γάρ qui est adverbial.
Notre corpus présente d ’autre part quatre emplois de άλλ' ή après négation, pour
lequel seule l ’interprétation adverbiale «autrement que» est acceptable en synchronie
(l’accentuation adjectivale άλλ' ή serait même parfois acceptable):

(1) ούδέν γάρ έστ' ά λλ' ή κοαξ

(«Carcen’estrien d ’autre que koax», 227; de même 928,1073, 1130)

Cependant des tours conjonctifs proches ont pu interférer (Denniston p. 26-27). Un

tel tour conjonctif est sans doute à l ’origine de l ’emploi de άλλα seul au sens
«excepté», selon une évolution comparable à celle du b u t anglais dans «no one but
you» (Denniston p. 3-4). Notre corpus en présente un exemple ignoré de Denniston
(qui, hormis un exemple de Sophocle, ne connaît cet emploi que chez Aristote):

(2) ού γάρ μούστίν ά λλ' άκουστέα ...

(«Je n ’ai pas d 'autre choix que d ’entendre...», 1380)

Cependant, en synchronie, cet emploi n’est plus conjonctif. Il équivaut à πλήν.

On peut aussi hésiter entre un emploi primitivement conjonctif et un emploi
adverbial dans le cas de l ’emploi apodotique de άλλα, absent de notre corpus. Selon
que la protase est considérée comme une indépendante ou une subordonnée, on
traduira en effet «mais» ou «en revanche» (voir Ruijgh 1971: 785). Mais, au moins
en synchronie attique, où la protase fonctionne comme une subordonnée, il paraît
préférable d’interpréter alors αλλά comme un adverbe.

2.2. Le cas des combinaisons commençant p a r où

Il est plus difficile de se déterminer en ce qui concerne les combinaisons ού μην

άλλα, ού μέντοι. άλλά, ού γάρ άλλά (Denniston ρ. 28-32). Ού μην άλλά
«néanmoins», ainsi que son doublet ού μέντοι. άλλά, est devenu un conglomérat
inanalysable en synchronie. Selon l’explication historique la plus courante, il s’agit
d’un tour où ce qui précède άλλά était à l’origine une phrase négative elliptique
(Denniston p. 28). Mais, selon Denniston, il faudrait souvent donner un contenu très
vague à cette phrase négative. D’autre part, le parallèle du français n é a n m o i n s , c’est-
à-dire à l’origine «en rien moins», suggère une explication différente. On peut en
effet imaginer une combinaison semblable à celle de n é a n m o i n s à l ’origine de ces
conglomérats: une combinaison où la négation aurait porté sur un adverbe et qui
aurait signifié «non autrement», et à partir de là «de toute façon», «pourtant» (cf.
français t o u t d e m ê m e ) .
Ces deux conglomérats ont une répartition très limitée dans la prose attique
(Denniston p. 29), et sont absents de notre corpus. Il n’en est pas de même de ού
γάρ άλλά, pour lequel, selon les exemples, les deux interprétations sont possibles ou
 Άλλ' έξόλοισθ’ αύτφ κοαξ 79

seulement la seconde. C’est ce que montre la comparaison de deux des trois exem­
ples des Grenouilles.

(3) HP. Ξυνεγένου τω Κλεισθένει;

ΔΓ. Μή σκώπτέ μ, ώ 'δελφ'· ού γάρ άλλ’ έχω κακώς·
(«—Tu as été avec Clisthène?
—Cesse de te moquer, mon frère, c a r je suis de to u te façon mal en point»,
57-58)

(4) Δΐ. έγώ δ’ εσομαί σοι σκευοφόρος έν τω μέρει.

ΞΑ. Φέρε δή ταχέως αϋτ'· ού γάρ άλλα πειστέον.
(«—Et moi je serai à mon tour ton porteur.
—Eh bien, prends vite ces paquets. C a r il me faut d e to u te façon obéir»,
497-498)

En effet, dans le premier cas une réponse d ’abord négative est à la rigueur possible:
«N on (je ne suis pas allé avec Clisthène), m ais je vais mal». Mais ce serait ne pas
tenir compte du fait que la réplique précédente est une raillerie, pas une vraie ques­
tion. Et rien ne suggère une réponse négative dans le second exemple, puisque
Xanthias acquiesce. Enfin, au vers 192, où l’on attend bien une réponse négative, elle
est en fait exprimée, avant ού γάρ αλλά, par M à τόν Δία. Il me paraît donc
préférable de préserver l ’unité de la combinaison ού γάρ άλλά en la traduisant
partout «car de toute façon», avec γάρ conjonction et ού ... άλλά adverbial (non
autrement > tout de même, de toutefaçon, pourtant).

3. P rélim in a ires théoriques

3.1. L a définition d ’u n « co o rd in a teu r binaire»

Une fois éliminés les emplois non conjonctifs de άλλά, il reste à préciser de quelle
sorte de conjonction il s’agit. Tout d ’abord, c’est un «coordinateur binaire», selon la
distinction de S.C. Dik (1968: 48), c’est-à-dire qu’elle coordonne deux membres et
pas plus. La description d’une coordination par άλλά consiste donc à décrire la rela­
tion P άλλά Q. Il faut pour cela:
1) définir la relation sémantique que άλλά établit entre P et Q,
2) définir la fonction syntaxique ou pragmatique de P et de Q, qui peut varier
selon les cas, mais est comparable pour P et pour Q, même si l’on observe de légères
distorsions (Dik 1968 etRuijgh 1971: 120),
3) définir la nature de l’ensemble P άλλά Q.
80 L o u is b a s s e t

3.2. Distinction de conjonctions sur critères morpho-syntaxiques

On a parfois distingué plusieurs sortes de conjonctions selon le statut syntaxique ou

morphologique des éléments qu’elles peuvent coordonner. Ainsi, on a opposé en
français, d ’une part o r et c a r , qui n’introduisent que des phrases, et d ’autre part
m a i s , o u , e t , n i qui peuvent introduire des phrases, des expressions ou des mots
(Wagner-Pinchon 1962: 422). Pourtant, l ’emploi de m a i s avant une simple expres­
sion ou un mot isolé est bien plus limité que celui de e t. H vaudrait donc mieux, même
selon ce critère, distinguer ces deux conjonctions.
Cependant, la différence entre e t et m a i s n’est pas liée au statut syntaxique ou
morphologique des éléments qu ’ils peuvent coordonner. Entre «Il est petit e t fort» et
«Il est petit, m a i s fort», la différence est d ’intonation, marquée par la virgule. Ceci
nous oriente vers une opposition d ’ordre pragmatique. Quand m a i s ou άλλα coor­
donnent des expressions ou des mots isolés, les éléments coordonnés sont tous deux
centres d’énoncé, ce qui n ’est pas toujours le cas avec e t («Pierre e t Jean sont petits»
n’a pas d’équivalent avec m a i s ) . Ceci suffit à montrer que les conjonctions adversa­
tives doivent être interprétées en ternies de pragmatique.

3.3. Classements d ’emplois sur critères morpho-syntaxiques

On est vite amené à la même conclusion lorsqu’on tente un classement des emplois
sur des critères distributionnels. Très vite ceux-ci se révèlent en effet inopérants.
Ainsi, s’agissant de classer les emplois conjonctifs de άλλα dans les G r e n o u ille s , on
est tenté de recourir aux critères suivants, qui en effet ont souvent été employés:
a. Opposition des emplois en début de réplique (P prononcé par un autre locu­
teur, on non verbal) et des emplois à l’intérieur d ’une réplique (même locuteur).
b. Opposition des emplois avec négation et des emplois sans négation, en distin­
guant selon que la négation apparaît avant ou après la conjonction (Denniston p. 1).
c. Opposition des emplois où P ou Q sont des phrases entières et de ceux où P
ou Q sont réduits à des expressions ou à des mots, avec des cas particuliers. Ainsi,
comme le suggère le cas de μάλλά, P peut être réduit à une négation. De même Q
peut se réduire à un mot ou groupe de mots.
On constate en effet que des emplois fréquents en début de réplique apparaissent
aussi à l ’intérieur de répliques, et inversement. De même ce n ’est pas n’importe
quelle négation qui est pertinente dans le second critère; et une expression ou un
simple mot peut fonctionner comme un énoncé autonome, aussi bien qu’une phrase
entière. Il faut donc reformuler les critères de classement ci-dessus en partant des
distinctions de la pragmatique.

3.4. Les distinctions sémantico-pragmatiques

O. Ducrot (1984: 13-46) a distingué trois composantes sémantiques présentes dans

un acte de parole. Mais la première de ces trois composantes se subdivisant en deux
 Άλλ' έξύλοισθ' αύτώ κοαξ 81

sous-composantes, il s’agit en fait de quatre composantes sémantiques. En effet, un

acte de parole étant la production d’un énoncé dans une situation d’énonciation que
cette production modifie en partie, on peut distinguer, d’une part ce qui n’est pas
modifié par l’acte de parole et la modification qu’il apporte, d’autre part ce qui est
sémantiquement inhérent à l’énoncé et ce qui relève seulement de sa production dans
une situation donnée:
a. Le présupposé correspond à ce qui dans l’énoncé est signifié à la fois comme
savoir préexistant commun au locuteur-énonciateur et au(x) destinataire(s) et comme
ne faisant pas l’objet de l’acte de communication.
b. En fait partie, ou plutôt s’y ajoute, le thème de discours préétabli entre eux, et
qui fonde pour eux ce qu’on a appelé Yisotopie du discours.* 5*Avec le présupposé, il
représente ce qui normalement n’est pas modifié, n’étant pas l ’objet de l’acte de
communication. En revanche, tandis que le présupposé est inhérent à l’énoncé, le
thème de discours et son isotopie sont déterminés par la situation d ’énonciation.
c. Le posé correspond à la modification statutaire apportée par l’acte de parole à
la situation d ’énonciation du destinataire. Elle est statutaire en ce que le posé est
inhérent au sens de l’énoncé produit, comme le présupposé. Tous deux constituent le
sens illocutoire de l ’énoncé.5
d. Le sous-entendu correspond à ce que le locuteur-énonciateur veut faire enten­
dre en produisant l’énoncé qu’il produit dans une situation d ’énonciation donnée.
C’est une modification non statutaire de la situation d ’énonciation. Cet élément du
sens, ni impliqué ni inhérent à l’énoncé, n’est atteint que par une induction: on dit y
(posé) dans une situation d ’énonciation donnée pour faire entendre aussi z (sous-
entendu). Il correspond aux intentions argumentatives du locuteur. Il s’agit du sens
perlocutoire de l’énoncé en situation.
Ce sens perlocutoire, avec les déterminations de l’isotopie, est ce qui permet de
distinguer l’objet abstrait qu’est un énoncé de Y acte de parole, son occurrence sin­
gulière. Les distinctions sémantico-pragmatiques proposées ici sont résumées dans le
tableau suivant.

élément non objet objet

sémantico-pragmatique de l’acte de parole d e l ’acte de parole
inhérent à l’énoncé PRÉSUPPOSÉ POSÉ
lié à l’énonciation THÈME DE DrSCOURS SOUS-ENTENDU

5 Groupe d’Entrevernes (1985:123): «L’isotopie garantit l’homogénéité d ’un message ou d ’un dis­
cours.»
5 II existe cependant un posé de simple reprise. Par exemple après puisque, le posé ayant été
préassetté.
 LOUIS BASSET

3.5. L es com posantes p ra g m a tiq u e s de l ’énoncé

Ces distinctions de sémantique pragmatique peuvent être complétées par une analyse
de l ’énoncé en termes de fonctions pragmatiques. L ’analyse la plus courante oppose
le thème (ou topique) , ce dont on parle, qui est censé connu, au rhème, ce qu’on en
dit, qui constitue le centre pragmatique de l’énoncé, l’apport nouveau. Le thème, qui
n’est pas toujours présent dans un énoncé, est supposé ou présupposé. Le rhème
contient souvent aussi du présupposé, mais il porte surtout le posé, ce qui le rend
pragmatiquement central et constitutif de l’énoncé.
La Functional Grammar de S.C. Dik fait cependant une analyse plus fine, à trois
termes: thème, topique et rhème. Dans cette tripartition, le thème est le thème du
discours, qui, à l ’intérieur d ’un discours ou d ’un dialogue, peut être commun à
plusieurs énoncés. Le topique en revanche est le thème de l'énoncé. Prenons pour
exemple le petit discours suivant:
«A Amsterdam, les canaux sont pleins de glaçons. H fait très froid. Mais les
vedettes circulent encore.»
Le thème de discours commun aux trois énoncés est Amsterdam aujourd’hui. Il
fonde une isotopie qui détermine ce que désignent ici les canaux et les vedettes, qui
sont deux thèmes à!énoncé différents appartenant à ce thème de discours. Le
deuxième énoncé, entièrement rhématique (ou thétique) n’a pas de thème.7 Ce petit
discours présuppose l’existence de canaux à Amsterdam, et que des vedettes y circu­
lent habituellement.
Le thème du discours, qui détermine une isotopie, est constitutif de la situation
d ’énonciation. On ne peut changer le thème du discours sans entrer dans une nou­
velle situation d’énonciation, ce qui rompt l ’nnité du discours ou du dialogue. En
revanche, des énoncés successifs, avec des thèmes d’énoncé différents pris dans la
situation d’énonciation, ne font qu’enrichir celle-ci de leurs apports nouveaux, qui
s’ajoutent au présupposé. L ’unité du discours ou du dialogue n’en est pas rompue.

3.6. A pplication a u x conjonctions adversatives

L ’unité d ’un discours ou d ’un dialogue n’est donc pas rompue tant que les actes de
paroles successifs n’invalident aucune des composantes sémantiques de l’acte de
parole qui les précède, se contentant d’apporter un posé et un sous-entendu nou­
veaux. La rupture de l’unité discursive correspondra soit à un changement de thème
de discours, soit à l ’invalidation du présupposé, du posé ou du sous-entendu de
l’acte de parole précédent.
Je fais l’hypothcsc qu’une conjonction adversative comme αλλά a pour fonction
d’introduire le rejet d ’une de ces quatre composantes sémantiques. Une rupture dis­
cursive est donc normalement marquée par son emploi. Mais il n ’est pas exclu
qu’elle puisse aussi être employée sans discontinuité discursive véritable, pour peu

7 II arrive cependant qu e thèm e d e discours e t thèm e d ’é noncé coïncident. Cf. D ik (1978: 141-142).
 Άλλ' έξόλοισθ' αύτφ κοαξ 83

que la composante sémantique invalidée n’ait pas été effectivement adoptée dans ce
qui piécède, mais soit restée virtuelle.
Sur ces bases théoriques, je propose une répartition pragmatique des emplois de
άλλά dans les G r e n o u i l l e s , en distinguant d ’abord ceux qui répondent à un compor­
tement, puis ceux qui marquent une rupture discursive, enfin ceux qui ne rompent pas
une continuité discursive.

4. Classement des emplois de ά λλά dans les G r e n o u i l l e s

4.1. La conjonction ά λλά répond à un comportement non verbal

Si la réplique introduite par άλλά répond à. un comportement non verbal, il n ’y a

dans ce qui précède ni présupposé, ni posé (sauf peut-être dans certains gestes
codés). Le seul des éléments sémantiques définis ci-dessus qui puisse apparaître est
un sous-entendu de situation «cela va continuer», et la conjonction peut servir à cor­
riger ce sous-entendu. Elle est alors une marque d’impatience. En l’absence d ’un tel
sous-entendu, elle indique seulement l'irruption d ’une nouvelle situation d’énoncia­
tion (cas particulier de l’emploi de découverte ou de surprise).
C ’est à l’emploi d ’impatience que correspondent les deux exemples des
G r e n o u i l l e s où Dionysos réagit au tintamarre que constitue le coassement des
grenouilles:5

(5) BA. Βρεκεκεκεξ κοαξ κοαξ.

ΔΙ. Ά λ λ’ έξόλοισθ’ αύτφ κοαξ
(«— Brékékékéx, coax, coax.
— Maïs puissiez-vous crever de votre coax même», 226; de même 240)

4.2. La conjonction ά λλά marque une rupture discursive

Généralement la réplique répond à du verbal. Il peut alors arriver que le nouveau

locuteur continue un propos précédemment amorcé (par le locuteur précédent ou par
lui-même avant une interruption). Il s’agit alors en fait d ’une continuation de
réplique. Mais s’il s’agit bien d’un propos nouveau, le nouveau locuteur se trouve
dans une situation d ’énonciation léguée par le propos précédent. La conjonction
adversative indique alors qu’il n’accepte pas cette situation d ’énonciation te lle q u e lle .
Le dialogue est alors continué au prix d ’une c o r r e c tio n affectant l ’une ou l ’autre des
quatre composantes sémantiques de l’acte de parole précédent.
Même à l’intérieur d’une réplique, la conjonction adversative peut indiquer une
rupture de l’unité discursive. Le locuteur peut instaurer par exemple un dialogue fictif
(figure de V h y p o p h o r e ou o c c u p a tio ) . B donne ainsi la parole à un interlocuteur fictif
dont la situation d’énonciation ne correspond pas à la sienne. Cette figure, qui a été
décrite pour άλλά par Denniston (I, 3, iii, p. 8-9), et pour le latin a t par C. Kroon
(1995: 340-2), semble cependant absente des G r e n o u ille s .
 LOUIS BASSET

M a i s c n d e h o r s m è m e d e c e t t e f i g u r e , u n e r u p t u r e d e l ’u n i t é d e d i s c o u r s r e s t e
p o s s i b l e . L e lo c u t e u r p e u t e n e f f e t s o it d é c i d e r d e c h a n g e r d e thème de discours, s o it
a v o i r f a i t e n p a r i a n t u n e d é c o u v e r t e q u i m o d i f i e s a s i t u a tio n d ’é n o n c i a t i o n ( d ’o ù c o r ­
r e c t i o n d e posé o u d e présuppose), s o it r e n o n c e r après-coup à un sous-entendu .

4 .2 .1 . L a c o n jo n c tio n ά λ λ α m a r q u e u n c h a n g e m e n t d e th è m e

E n d é b u t d e r é p liq u e , o n p e u t o p p o s e r u n n o u v e a u th è m e d e d is c o u r s a u p r é c é d e n t:

C6) HP· Μ ί α μ έ ν γ ά ρ έ σ τ ι ν a rc ò κ ά λ ω κ α ι θ ρ α ν ί ο υ ,
κ ρ εμ ά σ α ν τι σ α υτό ν.
Δ Ι. Π α ύ ε , π ν ιγ η ρ ά ν λ έ γ ε ις .
HP. Ά λ λ ’ έ σ τ ιν α τ ρ α π ό ς ξ υ ν τ ο μ ο ς τ ε τ ρ ιμ μ έ ν η , ή δ ι ά Θ υ ε ία ς .
(« — I l y a u n c h e m i n , q u i p a r t d ’u n e c o r d e e t d ’u n ta b o u r e t: i l s u f f i t q u e tu te
p en d es.
— A r r é te , c ’e s t d ’u n c h e m in tr o p r e s s e r r é q u e tu m e p a r ie s !
— A l o r s , i l y a u n s e n t i e r b a ttu , u n r a c c o u r c i . I l p a s s e p a r l a c o u p e à p o is o n » ,
123)

A D i o n y s o s q u i l u i a d e m a n d ò u n c h e m i n p o u r a l l e r d a n s l ’H a d è s , H é r a c l è s a p r o ­
p o s ti u n p r e m i e r i t i n é r a i r e , q u i e s t d o n e d e v e n u l e th è m e d u d i a l o g u e , m a i s r e j e t é
e n s u i t e p a r D io n y s o s . H é r a c l è s p r o p o s e a lo r s u n n o u v e l i tin é r a ir e , n o u v e a u th è m e d e
d i a l o g u e . C ’e s t l ’e m p l o i transiti/ oü progressi/ ( D e n n i s t o n I I , 9 , p . 2 1 ) , q u ’o n
o b s e rv e a u s s i d a n s d e s r é c its , p o u r p a s s e r à u n n o u v e l é p is o d e , s o u v e n t tra d u it p a r
alors o u eh bien.
E n d é b u t d e r é p l i q u e , l e c h a n g e m e n t d e th è m e d e d is c o u r s p e u t a u s s i c o r r e s p o n -
d r e à l ’in tr o d u c t i o n d ’u n p i a n d ’a c tio n , a p r è s u n c o n s t a t ( D e n n is to n I , 3 , ii, p . 8):

(7 ) A I. ’Ή δ η ’π ά τ α ξ α σ ';
ΞΑ. Ο ύ μ ά Δ ί' ο ύ δ α μ ο ι δ ο κ έ ϊς .
A I. Ά λ λ ' έ ίμ ' έ π ί το ν δ ί κ α ί πα τά ξω .
(« — T ’a i- je d é j à f r a p p é ?
— N o n , p a r Z e u s , n u l l e p a r t, a p p a r e m m e n t.
— E h b i e n j e v a is a l l e r f r a p p e r c e lu i- c i» , 6 4 6 , d e m è m e 5 7 7 )

A p rè s a v o ir so u m is X a n th ia s à la q u e s tio n , s a n s r é s u lta t, É a q u e d é c id e d e p a s s e r à
l ’é t a p e s u iv a n te , e n c h a n g e a n t d e v ic tim e .
M a i s c ’e s t l e p l u s s o u v e n t à l ’in t é r i e u r d ’u n e r é p l i q u e q u e l a d é c i s i o n d e p a s s e r à
F a c t i o n c o n s t i t u e u n e r u p t u r e d u th è m e d e d i s c o u r s p r é c é d e n t . O n p a s s e a i n s i d ’u n
c o n s t a t d ’é c h e c à u n e n o u v e l l e te n ta tiv e ( 6 6 2 , 6 6 9 , 1 1 0 3 , 1 2 3 5 , 1 2 4 8 , 1 3 9 7 , 1 4 3 5 ) , d u
m e n u à l ’i n v i t e ( 5 0 7 , 5 1 2 , 5 1 7 ) , d e s r c p r o c h e s à l ’e x h o r t a t i o n à m i e u x f a i r e ( 7 0 0 ,
7 3 4 ) , d e l ’o p i n i o n d ’u n t i e r s à c e l l e d e s i n t e r l o c u t e u r s ( 1 4 2 6 ) , e t m è m e d u t r o u b le
 Άλλ' έξόλοισθ' a m ß κοαξ 85

physique à la prescription médicale (!), selon un modèle relevé par Denniston (H, 9,
p. 21) chez Hippocrate.

(8) ΔΙ. Άλλ' ώρακιώ.

Ά λλ' οισε προς την καρδίαν μου σπογγιάν.
(«■—Mais je défaille. Eh bien passe moi une éponge sur le cœur», 482)

Dans d’autres cas s’exprime par la rupture discursive la volonté de réagir à une
situation fâcheuse (568, 843,1338,1356), ou de mettre fin à un bavardage ou à une
digression (108, 315, 377, 812, 905,1004).
Cela fait au total 27 emplois de αλλά que l ’on peut imputer à un changement
thématique, dont 24 à l’intérieur d’une réplique. Cet emploi dit t r a n s i t i f ou p r o g r e s s i f
est le moins correctif des emplois de άλλά. Il constitue une sorte d’emploi minimum.
On peut parfois imaginer un sous-entendu de l’énoncé précédent qu’il viendrait cor­
riger, en particulier lorsqu’il est accompagné d’une certaine impatience, par exemple
après la description d’une situation fâcheuse ou qui s’éternise (le sous entendu cor­
rigé serait alors: «Et cela va continuer»),

4.2.2. L a conjonction ά λλά m a rq u e u n e co rrection de présupposé

En début de réplique, un nouveau locuteur peut o b j e c t e r que ce qu’on vient de lui dire
repose sur des prémisses implicites inexactes (Denniston, I, 3, i, p. 7). Ces prémisses
peuvent être liées au lexique (par exemple, e n c o r e présuppose que ce qu’il modifie a
déjà existé), ou à la syntaxe (un syntagme nominal défini présuppose l’existence du
référent), ou simplement à la connaissance du monde.

(9) ΔΙ. Τούτων ε χεις ψέγειν τι;

ET. Πλεΐν ή δώδεκα.
ΔΙ. Ά λλ' ουδέ πάντα ταύτά γ’ εστ’ άλλ’ ή τρία.
(«— As-tu un reproche à faire à ces vers?
— Plus de douze!
—Mais ils ne sont en tout que trois», 1130)

Avoir plus de douze reproches à faire présuppose que le texte critiqué soit plus long
que trois vers. La réponse d’Euripide ne tient pas compte de cette donnée factuelle
première.
Dans d ’autres exemples, qui introduisent une réponse à un ordre ou à une
demande, on a affaire à ce qu’on a parfois appelé un «présupposé pragmatique». Le
locuteur indique que des conditions nécessaires pour justifier l’ordre ou la demande
ne sont pas remplies.

(10) ΔΙ. Ή δαιμόνιε, πρόσελθε· δέομαι γάρ τι σου.

HP. Ά λ λ’ ούχ οιός τ' εϊμ' άποσοβήσαι τόν γέλων.
 LOUIS BASSET

(«— Approche, diable d’homme! Car j ’ai besoin de toi.

— Mais je n’arrive pas à chasser ce rire», 45; de même 134,287, 292, 481)

Toute demande présuppose que l’interlocuteur est en mesure de l ’écouter. Ce n’est

pas ici le cas d’Héraclès qui détrompe Dionysos.

Un autre «présupposé pragmatique» de la demande ou de l’ordre est que l ’interlocu­

teur n ’est pas déjà décidé à faire ce qu’on lui demande (de même que le présupposé
d’un énoncé d’information est que le locuteur n ’est pas déjà informé). L ’emploi de
άλλά permet alors de corriger ce présupposé en contestant l’à-propos de l’ordre ou
de la demande, jugée superflue. C’est l’emploi d’assentiment impatient ou ironique,
qui correspond au français «Mais oui!»: «agreement is presented as self-evident and
inevitable» (Denniston Π, 6 et 7, p. 16-20).8

(11) ΔΙ. Οιμώζει;'· ού γάρ μοι μέλει,

ΒΑ. Άλλα μην κεκραξόμεσθά γ’
όπόσον ή φάρυξ civ ημών
χανδάνη δι' ήμέρας
(«—Gémissez! Je n ’en ai cure.
— Mais oui! Nous continuerons à pousser tous les cris dont notre gosier
sera capable, la journée endère», 258)

Les grenouilles n’ont pas besoin d’encouragements (qui sont d’ailleurs ici une ruse
pour les faire taire). Quoi qu’on dise, elles coasseront.

On dénombre donc 7 exemples où la conjonction άλλά introduit une correction de

présupposé. Cet emploi correspond à une correction assez forte. Il introduit une
o b j e c t i o n , sans cependant s’en prendre directement au contenu de ce qui a été dit.

4.2.3. La conjonction ά λ λ ά m arque une correction de posé

Il arrive que le nouveau locuteur s’en prenne directement au contenu de ce qui vient
d’être dit, parce qu ’il le juge insuffisant ou incorrect.
S’il le juge insuffisant, ou même douteux, il demande un complément d’infor­
mation (Denniston H, 1, i, p. 9).

(12) ΔΙ ...
Où tâv έτερός γ' αΰτ' είργάσατ’ άνήρ.
ΞΑ. Ά λ λά τί;
(«— Quelqu’un d’autre n’aurait certainement pas fait cela.

8 11 s ’agit de la variante (a), p. 16. La variante (b), p. 16, est une variante de l’emploi de découverte,
le locuteur découvrant, après hésitation, comme «malgré lui», son accord avec son interlocuteur. Ce
qui est corrigé est alors son attitude précédente de refus.
 'Αλλ' έξόλοισθ1αύτώ κοαξ 87

—Mais alors quoi?», 488)

Xanthias manifeste son incrédulité devant la prétention de Dionysos à avoir eu une

attitude exceptionnelle de bravoure.
Inversement, l’emploi de découverte, parfois avec άλλ’ η (Denniston IV, 4, p. 27),
permet d’apporter un complément d ’information, même à l’intérieur d’une réplique.

(13) Ξα. Τουτί τί ήν τό πράγμα;

Ά λλ’ ή Διός Κόρινθος έν τοΐς στρώμασιν;
(«— Qu’est-ce que c’est que çà? M ais c’est “Corinthos fils de Zeus’’ dans
le baluchon», 439, si on lit fj, conjecture de Bergk, alors que les manuscrits
ont ή)

Mais αλλά peut aussi introduire un désaccord sur le contenu, équivalant à notre
«Mais non!». Il sert alors à contredire.

(14) ΔΙ. ΈκΜ αραθώνοςή

πόθεν συνέλεξας Ιμονιοστρόφου μέλη;
AI. Ά λ λ’ ούν έγώ μέν είς τό καλόν έκ του καλοΰ
ήνεγκον αυθ’
(«— Es-tu allé ramasser des chants de puiseur d ’eau, à Marathon ou ailleurs?
—Mais non! Ils sont bien de moi, pris à noble source dans un noble but»,
1298, de même 1058)

Cet emploi c o n t r a d i c t o i r e est très fort et rare. Une variante moins forte apparaît dans
un dialogue croisé, où le locuteur substitue une suggestion à une première suggestion
qui, entre temps, a été rejetée par son interlocuteur.

(15) HP. Γυναικός;

ΔΙ. Où δήτ’.
HP. Ά λλά πα ιδός;
(«—D ’une femme?
— Certes non!
— Mais alors d ’un enfant?», 56, de même 57)

Ici, Héraclès, dans sa deuxième réplique, ne prolonge sans doute pas la réplique de
Dionysos, comme s’il achevait sa pensée (auquel cas l’allemand emploierait s o n ­
d e r n ) . Il corrige plutôt sa première suggestion, à la lumière de la réplique de
Dionysos (allemand; A b e r v i e l l e i c h t ...). Cette correction ne vise donc pas la dernière
réplique, mais l’avant-demière, quelque chose s’étant produit dans l’intervalle, qui
rend cette précédente suggestion caduque.
De même, à l ’intérieur d’une réplique, l ’emploi de découverte peut permettre de
contredire ce qu’on vient de dire.
 LOUIS BASSET

(16) AI.
ος τον κύν ήμών έξελάσας τόν Κέρβερον
άπήξας αγχών κάποδράς φχου λαβών,
... ’Αλλά νυν έχει μέσος·
(«toi qui as poursuivi notre chien Cerbère, t’es jeté sur lui en le serrant à la
gorge, et t’es enfui au loin en l’emportant... Mais maintenant, te voilà pris, en
plein dedans», 469)

On peut penser qu’Éaque, après s’être indigné du larcin d’autrefois resté impuni,
réalise soudain le renversement de situation: le voleur est revenu se jeter dans le
piège.

On a donc 7 exemples (6, si l ’on n’admet pas la correction de Bergk au vers 439) où
αλλά introduit une correction de posé, dont 5 sont contradictoires, emploi très forte­
ment correctif. Sur ces 5 exemples cependant, il en est 3 où la correction est préparée
par un événement intercalaire (découverte du locuteur ou réplique d’un interlocuteur)
qui rend caduc l ’énoncé contredit. On verra que ces emplois préparés peuvent être à
l’origine des emplois de αλλά correspondant à l’allemand s o n d e r n .

4.2.4. La conjonction ά λλά m arque une correction de sous-entendu

Un nouveau locuteur, tout en reconnaissant la vérité de ce qu’on lui a dit, peut cor­
riger un sous-entendu de son interlocuteur. Ce que traduit notre «Oui, mais ...»
(Denniston I, 3, i, p. 7).

(17) ΔΙ. Και μήν ού Παντακλέα γε

έδίδαξεν ...
AI. Ά λλ’ άλλους τοι πολλούς άγαθούς ...
(«— Eten tout cas il n’apas instruit Pantaclès...
— Oui, mais beaucoup d ’autres, des gens de bien ...», 1039; de même 136,
1025,1136,1215,1389)

Le sous-entendu de l ’énoncé de Dionysos est que l ’enseignement d ’Homère n’a pas

servi à grand-chose. C’est ce sous entendu effectif que corrige la réponse d’Eschyle.
Le locuteur peut même corriger un sous-entendu de son propre énoncé précé­
dent.

(18) XO. Μόνον όπως

μή σ’ ό θυμός άρπάσας
εκτός οΐσει των έλαών
δεινά γάρ κατηγόρηκεν.
Ά λλ' όπως, ώ γεννάδα,
μή προς οργήν άντιλέξεις.
 Άλλ' έξόλοισθ' αύτώ κοαξ 89

(«Prends garde seulement que ton ardeur ne t’emporte au-delà des oliviers.
Car terribles sont ses accusations. M ais prends garde, noble coeur, à ne pas
répondre sous l’empire de la colère», 997; de même 43,1046)

Cependant, pour que αλλά introduise ici une rupture discursive, il faut que l’énoncé
précédent «Car terribles sont ses accusations», constitue un acte de parole complet,
sous-entendant effectivement «tu as le droit de te fâcher». C’est bien ce que montre
le fait que γάρ ne porte que sur cet énoncé. La dureté des accusations d’Euripide
explique à elle seule la crainte exprimée auparavant. C’est un acte de parole justifi­
catif autonome avec son sous-entendu effectif, corrigé ensuite par un retour au pre­
mier acte de parole. On a donc trois actes de parole successifs:
1) «Ne te mets pas en colère».
2) «Je le dis car ses injures t’y autorisent».
3) «Mais évite quand même de te mettre en colère».
La portée limitée de καίτοι au vers 43 indique le même découpage en actes de
parole distincts. Ce découpage semble convenir aussi au vers 1046. Ce sont donc 3
exemples de rupture discursive reposant sur une correction de sous-entendu effectif à
l ’intérieur d ’un discours, ce qui fait 9 exemples avec ceux qui sont en début de
réplique.

En revanche, dans un énoncé comme «il t’a certes injurié, mais ne te fâche pas»,
certes, rendant le premier énoncé concessif, invalide d’emblée son sous-entendu «tu
peux te fâcher». Cet énoncé ne constitue donc plus un acte de parole autonome,
comme le montre la succession suivante: «Ne te mets pas en colère. Car, il t’a certes
injurié, mais la colère est mauvaise conseillère». Car porte alors sur tout ce qui suit,
et même plus précisément sur la partie introduite par mais. Cette portée large de car
a pour effet de réunir les deux énoncés en un seul acte de parole. Il n ’y a plus que
deux actes de paroles successifs:
1) «Ne te mets pas en colère».
2) «Évite-le car (certes ses injures pourraient t’y autoriser, mais c’est décon­
seillé)».
Nous voyons pat cet exemple fictif comment mais, à l ’intérieur d’un discours,
peut développer un emploi où il n’introduit plus un nouvel acte de parole avec rupture
discursive. Il suffit pour cela que l’énoncé qui précède la conjonction annonce et soit
sémantiquement suborbonné à celui qu’elle introduit, son sous-entendu virtuel ne
devenant pas effectif dans un acte de parole autonome. On observe des emplois
semblables de la conjonction grecque αλλά.

4.3. α λλά ne m arque pas une rupture d’unité discursive

Pour qu’une conjonction adversative à l ’intérieur d’une réplique introduise une cor­
rection de ce qui précède sans pour autant rompre l’unité du discours, il faut que la
composante sémantique qu’elle invalide n’ait pas été assumée par le locuteur, qu’elle
 LOUIS BASSET

soil donc demeurée virtuelle. Or ceci ne peut se produire que pour ce qui constituerait
un apport nouveau de l ’énoncé, c’est-à-dire sous-entendu ou posé. Thème de dis­
cours et présupposé sont en effet donnés avant même la prise de parole.
La correction peut affecter un sous-entendu resté virtuel dans les cas analogues à
celui qui vient d’être évoqué pour mais. Le locuteur accomplit alors un seul acte de
parole en liant deux énoncés successifs pour produire un seul sous-entendu. Comme
le sous-entendu virtuel du premier énoncé est contredit dans le second énoncé, ce
sous-entendu est d'emblée invalidé par l’intention du locuteur d’enchaîner les deux
énoncés en un seul acte de parole: «H est petit, mais fort». Il s’agit en ce cas de deux
énoncés décrivant deux situations référentielles différentes, le second énoncé invali­
dant un sous-entendu virtuel du premier (par exemple «Il est faible»).
Il peut arriver aussi que la correction ne vise pas un sous-entendu qu’on pourrait
tirer de l ’énoncé précédent, mais un sous-entendu qu’on pourrait inférer de la seule
situation de communication. En ce cas, les deux énoncés successifs peuvent avoir des
contenus et des sous-entendus semblables.
La substitution peut enfin affecter un posé resté virtuel. Mais ce n ’est possible,
sans palinodie, donc sans rupture de l ’unité du discours, que si un rhème a été l ’objet
d’un rejet dans ce qui précède, c’est-à-dire si son posé a été mis sous la portée d ’une
négation polémique. C’est donc seulement en corrélation avec une négation de rejet
qu’une conjonction adversative peut substituer un rhème à un autre, dans un seul acte
de parole: «Il n’est pas grand, mais petit». La même situation référentielle est cette
fois décrite sous deux formes différentes. On a suggéré ci-dessus (4.2.3.) que ce
rejet préalable a pu être à l’origine produit par un événement intercalaire suscitant un
άλλά de rupture discursive contradictoire.

4.3.1. ά λ λ ά corrige un sous-entendu virtuel d ’énoncé

Il n’est pas toujours facile de décider si l’énoncé qui précède άλλά est sémantique­
ment subordonné, avec une valeur plus ou moins nettement concessive, à celui qu’in­
troduit άλλά. La présence de οίδ’ ö u peut lui donner valeur concessive (v. 586, 601).
De même dans une réponse à une question orientée, quand un locuteur concède ce
qu’on veut lui faire dire (v. 1053). Mais en l’absence d’intonation le mouvement de
pensée du locuteur peut rester ambigu.
On a cependant vu ci-dessus (4.2.4.) que la portée d’un connecteur placé devant
le premier énoncé peut être un critère, selon qu’il porte seulement sur le premier
énoncé ou sur les deux. 11 arrive même qu’on ait ainsi un άλλά initial portant sur un
ensemble de deux énoncés séparés par un autre άλλά, selon la structure: Mais (A
mais B).19

(19) ΔΙ. Τουττ μέν σοι κακόν εϊργασταί' Θηβαίους γάρ πεπόηκας
άνδρειοτέρους εις ιόν πόλεμον· καί τούτου γ' οΰνεκα τύπτου.
AI. Ά λλ' ύμίν αϋτ' έξην άσκεΐν, ά λλ' ούκ έπί τοΰτ’ έτράπεσθε.
 Ά λλ' έξύλοισΟ’ αύτφ κοαξ 91

(«— En quoi tu as mal agi. C ar tu as rendu les Thébains plus courageux à la

guerre. Et pour cela reçois des coups.
— Mais vous pouviez vous exercer, mais n’y songiez pas», 1025)

Le premier άλλά, en tête de réplique, marque une rupture discursive. Il sert à corriger
le sous-entendu effectif «tu es fautif» de «tu as aguerri les Thébains». Or ce sous-
entendu n’est pas corrigé seulement par le premier énoncé introduit, mais par
l ’ensemble des deux énoncés qui sous-entend «c’est vous les fautifs».9
Le second άλλά, ainsi mis sous la portée du premier, ne peut corriger, lui, qu’un
sous-entendu virtuel du premier énoncé. Or, hors de toute situation particulière pro­
duisant le sous-entendu effectif contraire,10 la r é a lis a tio n est le sous-entendu virtuel
d e là p o s s i b i l i t é . 11 Ainsi, «vous n’y songiez pas» corrige directement par son posé,
un sous-entendu virtuel de l’énoncé immédiatement précédent («vous vous êtes exer­
cés»), et avec lui, mais indirectement, par un sous-entendu effectif («c’est vous les
fautifs, donc pas moi»), le sous-entendu effectif de la réplique précédente. La struc­
ture de l’ensemble est donc:
1) «Tu as aguerri les Thébains», ce qui sous-entend «tu es fautif».
2) M a i s («vous pouviez vous exercer», ce qui pourrait sous-entendre que vous
l ’avez fait, m a i s «vous ne l’avez pas fait»), ensemble qui sous-entend «c’est vous les
fautifs, et non moi».

Outre les cas où, comme dans l ’exemple ci-dessus, l ’énoncé introduit par άλλά
invalide directement, par son p o s é , un sous-entendu virtuel de l ’énoncé précédent
(voir aussi v. 601), c ’est parfois seulement un s o u s - e n t e n d u effectif de l ’énoncé
introduit par άλλά qui invalide, indirectement, le sous-entendu virtuel de l ’énoncé
précédent.

(20) ΞΑ. Où κακώς ... παροανεΐτ’,

ά λλά καυτός τυγχάνω ταΰτ
άρτι συννοούμενος.
(«— Votre conseil n ’est pas mauvais, mais moi-même j ’y songeais à
l ’instant», 598; de même 586 et 1053 pour le second άλλά du vers)

Le premier énoncé pourrait sous-entendre que le conseil donné est utile à Xanthias, le
second invalide ce sous-entendu indirectement, en s o u s - e n t e n d a n t qu’il lui est inutile.
Il est clair que Xanthias n ’a pas voulu d ’abord sous-entendre effectivement qu’il lui
est utile. Il ne prononce donc le premier énoncé que dans la perspective du second
qui invalidera son sous-entendu virtuel.

9 II est vrai que «Mais vous pouviez vous exercer!» serait une objection suffisante, mais seulement
dans la situation particulière où cet acte de parole sous-entend «vous ne l’avez pas fait».
10 Voir note précédente. Je distingue ici le sous-entendu effectif d ’un acte de parole et le sous-
entendu virtuel de l’énoncé produit. Ils peuvent s’opposer.
11 C ’est pourquoi, alors que la succession «vous pouviez le faire, mais vous ne l ’avez pas fait» est
naturelle, il n ’est guère naturel de dire «vous pouviez le faire, mais vous l’avez fait».
92 LOUIS BASSET

4.3.2. ά λλ' ού corrige un sous-entendu virtuel de situation

À ces emplois se rattachent certains emplois de άλλ' où. Ceux-ci, pourtant, au con­
traire de ce qui se passe pour l’exemple (19) ci-dcssus, semblent décrire le même état
de fait que l ’énoncé précédent, sous forme négative. C’est ce qui avait conduit Den-
niston à joindre ces emplois aux emplois «éliminatifs» de où ..., άλλα qui décrivent
aussi deux fois le même état de fait, mais d’abord de façon négative, et ensuite de
façon positive (1,1, i, a, p. 1-2).

(21) AI. ... ήμή φρήν ... πολλάς άρετάς έπόησεν,

Πατρόκλων, Τεύκρων ...

Ά λλ' ού μά Δί' ού Φαίδρας έποίουν πόρνας ούδέ Σθενεβοίας

(«... mon génie a mis en scène les nombreuses vertus des Patrocles, des
Teucros...
M ais par Zeus, je ne mettais pas en scène des Phèdres prostituées, ni des
Sthénébées ...», 1043; de même 961)

C.J. Ruijgh (1971: 136) a cependant fait remarquer que cet emploi «éliminatif
inverse» est traduit en allemand par a b e r , contrairement à l’emploi «éliminatif» qui
est traduit par s o n d e r n , ce qui suggère pour le moins une différence de valeur.
D ’autre part, si dans l’emploi «éliminatif», il y a bien substitution grâce à άλλα du
vrai au faux (le faux étant ce qui esc nié dans le premier énoncé), il n’en est rien dans
l’emploi «éliminatif inverse». Dans le premier énoncé, qui est alors positif, rien n’est
faux ni éliminé par ce qui suit. Et l ’énoncé qui suit άλλά est aussi présenté comme
vrai. C’est seulement à l’intérieur de cet énoncé que le «faux est éliminé», mais alors
par la négation polémique.
Il faut donc trouver ailleurs la justification de άλλά. La difficulté vient de ce que
les deux énoncés ne semblent se contredire ni directement par leurs contenus, ni indi­
rectement par leurs sous-entendus. On ne voit pas ce qui oppose «j’ai mis en scène
des Patrocles» et «je n’ai pas mis en scène des Phèdres». D ’ailleurs on observe la
concurrence de καί où.
La solution me semble être dans le fait qu’un sous-entendu virtuel peut naître
non seulement de l’énoncé, mais aussi de la situation d’énonciation. Dans l ’exemple
ci-dessus, Eschyle, parlant des héros de ses tragédies, en présence d ’Euripide, est
dans une situation d ’énonciation qui sous-entend qu’un poète tragique, donc lui-
même, peut mettre en scène des Phèdres et des Sthénébées. C’est ce sous-entendu
virtuel de situation, qui est peut-être même ce que sous-entendent ses interlocuteurs,
que veut corriger Eschyle, un sous-entendu qui n’appartient en propre ni à cc qu’il
vient de dire, ni à lui-même. C ’est pourquoi la négation est nettement polémique.
Mais un tel sous-entendu de situation n’est pas toujours perçu par le locuteur. C’est
ce qui explique la concurrence de καί où là même où l ’on attendrait άλλ’ où
(Denniston 1,1, p. 2).
 Άλλ’ έξόλοισθ' aimS κοαξ 93

On a donc dans les Grenouilles 7 exemples où άλλα corrige un sous-entendu

virtuel. Dans 5 exemples, c’est le sous-entendu d’un énoncé précédemment concédé
par le locuteur. Dans 2 exemples, c’est un sous-entendu de la situation. Mais il s’agit
dans tous les cas d ’un sous-entendu qui n’est pas le fait du locuteur dans l ’énoncé
précédent, c’est pourquoi il demeure virtuel.

4.3.3. L a conjonction ά λ λ α c o n tre d it u n posé virtuel

On a observé ci-dessus (4.2.3.) des emplois de rupture discursive contradictoire. Ils

pouvaient être suscités par un événement intercalaire intervenant entre l ’énoncé
contredit et l ’énoncé introduit par αλλά. Cet événement intercalaire pouvait être,
comme dans l ’exemple (15), l ’acte de parole d ’un interlocuteur rejetant, avec une
négation polémique, l’énoncé précédent du locuteur.
Si cependant les rôles sont inversés, si c ’est le locuteur qui rejette un énoncé de
l ’interlocuteur (posé ou seulement sous-entendu), ce rejet est naturellement intégré
dans son propre acte de parole. Le posé de l’énoncé contredit y est alors exprimé
comme n’étant pas le fait du locuteur, donc virtuel.
Ce mouvement de pensée peut être exprimé par deux énoncés complets, le pre­
mier de rejet, le second de substitution (Denniston 1 ,1, i, b, p. 1).

(22) ET. Ειτ' ούκ έλήρουν ö τι τύχοιμ1ούδ’ έμπεσοιν εφυρον,

άλλ’ ούξιών πρώτιστα μέν μοι τό γένος εΐπ’ αν ευθύς
τού δράματος.
(«Puis je ne divaguais pas au hasard ni ne fonçais dans le brouillard, mais
mon premier personnage entré en scène exposait aussitôt l ’origine de
l’action», 946; de même 949)

Cependant le thème des deux énoncés est commun. C’est ici la méthode dramatique
d ’Euripide, représenté par la première personne dans le premier énoncé, par μοι dans
le second. La correction ne porte en effet que sur un posé virtuel, contenu dans un
rhème rejeté, auquel est substitué après άλλά un rhème contradictoire. Mais le thème
ne change pas.
Ceci explique que les deux énoncés aient le plus souvent un sujet commun, ce
sujet représentant habituellement leur thème commun:

(23) AI. Και μην μά τόν Δί' ού κατ’ έπος γ έ σου κνί,σω τό ρημ' έκαστον,
ά λλα συν τοισιν θεοΐς άπό ληκυθίου σου τούς προλόγους
διαψθερώ.
(«— Eh bien, par Zeus, je n’irai pas gratter chacune de tes expressions, vers
par vers, mais, avec l’aide des dieux, j ’empoisonnerai tes prologues grâce à un
petit flacon», 1199; de même 360, 436, 696, 939, 999, 1011, 1117, 1376,
1407)
 LOUIS BASSET

Si le sujet du premier énoncé est une négation quantitative, celui du second est
indéfini. Les deux énoncés sont uniquement Thématiques:

(24) AI. Οϋκουν έθέλει γε τριηραρχείν πλουτών ούδείς διά ταΰτα,

α λλά ρακίοις περιειλάμενος κλάει καί φησί πένεσΟαι.
(«— C’est pourquoi pas un riche ne consent à être triérarque, m ais on se
couvre de haillons, on pleure et on prétend être pauvre», 1066)

Le fait que les deux énoncés n ’aient jamais de thèmes d ’énoncé distincts, est un
indice de leur interdépendance. Un indice encore plus net est l ’emploi du ού interro­
gatif d’impatience («ne vas-tu pas ...?», avec indicatif futur) portant sur l ’ensemble
des deux énoncés, et suivi d’un μή écartant préalablement le premier rhème, selon la
structure (ού (μή A, άλλά B)):

(25) XA. Ού μή φλυαρήσεις έχων, άλλ’ άντιβάς έλ$ς προθύμως;

«— Ne vas-tu pas ne pas t’obstiner à dire des bêtises, m ais prendre tes
appuis et pousser avec ardeur?», 202; de même 462, 525)

Il y a 16 exemples de ce type avec des verbes différents dans les deux énoncés. Dans
d ’autres, le verbe est commun aux deux énoncés, le rhème étant limité à un de ses
compléments, par exemple à un complément prédicatif. Le verbe n’est alors exprimé
qu’une fois, l’autre énoncé étant elliptique.

(26) ΞΑ. Άλλ' ούκέτ’ αΰ γυνή 'στιν, άλλ' ήδη κύων,

(«— Mais ce n ’est plus une femme, mais désormais une chienne», 292; de
même 722,725,1016,1161)

Parfois le verbe n’apparaît que dans le second énoncé. Comme dans l ’exemple (26),
il ne porte pas le contenu informatif de l’énoncé et a donc un rôle de thème. Les deux
énoncés opposent deux compléments possibles de ce verbe. Ce sont eux qui ont le
rôle de thèmes.

(27) ΔΙ. Ού τάχ', ά λλ’ήδη ποιώ.

(«— Ce n’est pas bientôt, mais maintenant que je le fais», 527; de même 117,
856, 1144))

On peut même ne pas avoir de verbe du tout, quand on a affaire à deux appositions
prédicatives (v. 970).
L ’ellipse du verbe après une mention précédente n ’est pas propre à cet emploi.
Même quand il y a rupture d’unité discursive, elle peut apparaître dans l ’énoncé qui
suit άλλά, comme dans l ’exemple (17), sauf s’il s’agit d’un changement thématique
(car en ce cas il ne peut y avoir reprise d’une mention précédente).
 Ά λλ' έξόλοισθ’ αύτφ κοαξ 95

En revanche, toute ellipse dans l ’énoncé qui précède άλλά et annonçant une
mention ultérieure, n’est possible que lorsque les deux énoncés sont sémantiquement
subordonnés l’un à l ’autre, c’est-à-dire lorsqu’il n’y a pas rupture discursive, mais
un seul acte de parole. Ce que corrige άλλά est alors virtuel, qu’il s’agisse d ’un
s o u s - e n t e n d u ou d’un p o s é , comme dans l’exemple (27).
Cette ellipse d’un verbe non Thématique apparaît dans 10 exemples. Mais le
premier énoncé peut même faire l’ellipse du r h è m e , c’est-à-dire du p o s é contredit, si
celui-ci est déjà exprimé par l ’interlocuteur, par exemple dans une question. Π arrive
donc que seuL le thème soit réexprimé.

(28) ΞΑ. Κάπειτα πώς

où και Σοφοκλέης άντελάβετο τοΰ θρόνου;
ΟΙ. Μ ά Δί’ ούκ εκείνος, άλλ' εκυσε μεν Αισχύλον,
ότε δή κ ατηλθε...
(«—Mais alors comment se fait-il que Sophocle ne se soit pas emparé du
trône?
—Non par Zeus, pas lui, mais il a embrassé Eschyle, lorsqu’il est descendu
ici», 788; de même 1457,1462)

Il y a ellipse du posé rejeté «il s’est emparé du trône», mais reprise du thème «lui».
Le tour est cependant bien plus fréquent sans aucune reprise, avec seulement l’ex­
pression du rejet. Il est remarquable que ce rejet ne soit alors jamais exprimé par un
simple où, mais par μή (103, 167, 611, 745, 751; cf. Denniston I, 1, iii, d, p. 4), ou par
μά (τον) Δία (174, 501, 663, 753,779,1053), rarement renforcé par où (650,1183).

(29) NE. Δύο δραχμάς μισθόν τελείς;

ΔΙ. Μ ά Δί', ά λλ’ έλαττον.
(«—Tu paieras deux drachmes de salaire?
—Non par Zeus, mais moins», 174)

Le rejet n ’est pas toujours un rejet du contenu exprimé, mais est parfois rejet de la
forme, c’est-à-dire d ’un mot employé par l ’interlocuteur. C’est par exemple pour
surenchérir.

(30) HP, Σε δέ ταυτ' άρέσκει;

ΔΙ. Μ άλλά πλείν ή μαίνομαι.
«—Et ces choses-là te plaisent?
— Ne dis pas cela, mais que j ’en suis plus que fou», 103; de même 753,
avec μά Δία)

Ce tour assez fréquent (16 exemples) permet de préparer la correction contradictoire

introduite par άλλά. Or on a vu, dans l’exemple (14), un emploi contradictoire de
άλλά sans ce rejet préalable du locuteur, où donc άλλά initie une rupture discursive
 LOUIS BASSET

très forte. On comprend que le tour alt élé le plus souvent prepare par l’expression
préalable du rejet. Le tour: — Δύο δραχμάς τελείς; — Ά λλ' έλαττον. «— Tu
paieras deux diachmes? — M ais non, moins!» a élé renforcé par Γ insertion de Mà
Δί', qui reqoit, en Tòtani à άλλα, le ròle d’initiateur d ’une rupture discursive, tout en
virtualisant le posé que corrige αλλά.
Μάλλά ou Μα Δί' άλλά sont done ici des variantes renforcées du άλλα de rup­
ture discursive contradictoire. Ces tours font done le lien entre les emplois de rupture
discursive, contradictoires et autres, et Temploi où άλλά après négation équivaut à
Tallemand sondern.12

5. C onclusion

5.1. R é p a rtitio n des em plois

Les exemples ont été répartis comme suit entre entre les trois grands groupes définis:
Λ. Discontinuité discursive: 52 (ou 51) exemples.
— Changemcnt de comportement (impatience): 2 exemples.
— Changement thématique: 27 exemples.
— Correction de présupposé: 7 exemples.
— Correction de posé: 7 (ou 6) exemples.
— Correction de sous-enlendu: 9 exemples.

B. Correction d’un sous-entendu virtuel: 7 exemples.

— Sous-entendu virtuel de l’énoncé: 5 exemples.
— Sous-entendu virtuel de situation (άλλ’ ού): 2 exemples.

C. Correction d’un posé virtuel (all. s o n d e r n ): 42 exemples.

— Sans ellipse: 16 exemples.
— Avec ellipse d’un verbe non rhématique: 10 exemples.
— Avec ellipse du rhème rejeté (jamais ού seul): 16 exemples.

Le classement de certains exemples à l’intérieur du premier groupe peut ótre contcsté.

La recherche de sous-entendus est ce qui se prète le plus aux interpolations subjec-
tives, comme Tont montré les travaux de Ducrot e t a l . On a vu que certains emplois
expliqués ici par un changement thématique pouvaient Tètre aussi par une correction
de sous-entendu. J’ai considéré que là où un autre facteur, plus objectif pouvait*1

12 Un tel lien a déjà été proposé par Ruijgh (1971: 136-7), qui cependanl donnait la prioriti au
type sondern. Cette conjonction particulière de Tallemand réunit deux caractéristiques qui la
rendeni doublement marquée :
1) Elle a un sens proprement contradictoire (correction de posé seulement).
2) Elle intervicnt seulement à Tintérieur d’un acte de parole (pas de rupture discursive).
Chacunc de ces deux caractéristiques se retrouve, mais isolément, dans des emplois du type aber.
 Ά λλ’ έξόλοισθ' αύτφ κοαξ 97

expliquer l ’emploi de la conjonction, ce facteur avait plus de force que l ’éventuel

présence de sous-entendus.
Malgré ces cas douteux, il ressort de notre examen que sur 101 (ou 100) occur­
rences de άλλα conjonctif dans les Grenouilles, 52 (ou 51) signalent une rupture
d’unité discursive. Ces 52 occurrences constituent une petite majorité, au moins rela­
tive, face aux 7 occurrences qui corrigent un sous-entendu virtuel, et aux 42 qui corri­
gent un posé virtuel après négation de rejet (y compris les 16 emplois après négation
seule, en quelque sorte intermédiaires).
On voit, par rapport aux thèses de O. Ducrot et al., le changement de perspective
apporté par cette étude. Ce qui était pour eux le modèle standard (correction de sous-
entendu, du type «Il est petit, mais fort») est ici un emploi assez rare, alors que des
emplois considérés par eux comme marginaux ou d ’autres exclus de leur étude se
révèlent ici bien plus fréquents (changement thématique, correction de posé du type
sondern). En outre, ce changement de perspective n ’est pas vraiment dû au choix
d’un dialogue comme texte d ’étude. Sur les 52 exemples de rupture discursive, 29,
plus de la moitié, sont à l ’intérieur d’une réplique. Ces emplois sont donc loin d’être
spécifiques du dialogue.
Le point de vue adopté ici pennet d’intégrer à la description l’emploi du type
sondern. Cet emploi est certes fréquent dans les Grenouilles, mais non majoritaire
comme il le serait sans doute dans un texte sans dialogue.13 Si, malgré cette
fréquence, plus ou moins grande selon le type de texte, je ne le considère pas comme
central, c ’est que j ’admets l’unité sémantique de άλλά. L ’emploi du type sondern en
effet a une définition sémantique et fonctionnelle trop étroite pour couvrir, en syn-
clironie, les autres emplois de άλλά. Les divers emplois du type aber en revanche
supposent une définition sémantique et fonctionnelle plus souple. Celle-ci peut justi­
fier aussi les emplois du type sondern dans les langues qui n’ont pas de conjonction
spécifique pour ces emplois.

5.2. Definition de la coordination par ά λ λ ά

Voici comment ces divers emplois permettent de définir la coordination binaire

exprimée par άλλά, selon les trois exigences indiquées ci-dessus en 3.1.
— 1) Relation sémantique entre P et Q:
Dans la structure P άλλά Q, άλλά indique que Q corrige P. Une composante
sémantique de Q est substituée à une composante sémantique de P. Cette définition
sémantique unitaire de άλλά en fait une conjonction plutôt corrective qu’adversative
(la relation interlocutorale ne me paraît pas fondamentale, bien qu’elle soit souvent
présente, et même prépondérante dans le cas d’un sous-entendu virtuel de situation).
— 2) Fonctions pragmatiques de P et Q:

13 Ruijgh (1971: 135) évoque les comptes du lexique de Powell concernant Hérodote, avec 399
exemples de ούκ A άλλά B sur 515 de άλλά. Même si la présence d’une négation dans le premier
énoncé ne suffit pas pour définir l ’emploi comme «éliminatif», cette statistique montre que l’emploi
«éliminatif» est nettement majoritaire dans un récit.
 LOUIS BASSET

Dans la structure P άλλα Q, P et Q peuvent ètte:

a) deux actes de paroles autonomes, avec chacun quatte composantes sémantiques:
thème de discours, présupposé, posé, sous-entendu (mais P peut n’étre qu’une situa­
tion, avec seulement un sous-entendu). La correction peut alors porter sur n ’importe
laquelle des composantes sémantiques de P.
b) deux énoncés produisant un seul sous-entendu. La correction est limitée à un
sous-entendu virtuel de P, si cette orientation argumentative est contredite en Q (type
«Il est petit, mais fort»), ou à un sous-entendu inférable ni de P, ni de Q, mais de la
situation (type «J’ai mis en scène des Patrocles, mais pas des Phèdres»).
c) deux rhèmes d ’énoncés contradictoires (en ce cas P est sous la portée d’une néga-
tion polémique). La correction est limitée au posé de P (type «Il n ’est pas grand,
mais petit», emploi de l’allemand sondern).
— 3) L ’ensemble P άλλα Q constitue:
dans le cas a) un dialogue ou un discours avec discontinuité discursive,
dans le cas b) un seul acte de parole liant deux énoncés pour produire un seul sous-
entendu.
Dans le cas c), l’ensemble ού P άλλά Q constitue un groupe rhématique binane avec
un seul thème d’énoncé.
La variété des emplois de αλλά n ’est done pas la conséquence de variations
sémantiques qui lui seraient propres, mais des divers objets que cette conjonction
relie et des diverses composantes sémantiques qu’elle corrige.

B ib lio g ra p h ie

Anscombre, A J.C. et O. Ducrot

1977 ‘Deux mais en ffanqais’. Lingua 43,23-40
Basset, L.
1988 ‘Valeurs et emplois de la particule dite modale en grec ancien’. In: In the
Footsteps o f Raphael Kühner. Amsterdam , 27-37
Chanet, A.-Μ.
1990 ‘Négation sur parataxe et structures apparentées en grec: comment se con­
struit rinterprétation globale’. Information Grammaticale 46, 28-33.
Denniston. J.D.
1954 The Greek Particles. Oxford Clarendon Press (2e édition)
Dik, S.C.
1968 Coordination, its Implications for the Theory o f General Linguistics.
Amsterdam
1978 Functional Grammar. Amsterdam
Ducrot, O.
1972 Dire et nepas dire. Principes de sémantique linguistique. Paris: Hermann
1984 Le dire et le dit. Paris: Minuit
 Άλλ’ έξόλοισθ' αύτφ κοαξ 99

Ducrot, Ο. et al.
1980 Les mots du discours. Paris: Minuit
Groupe d’Entrevemes
1985 Analyse sémiotique des textes. Lyon: PUL (5e édition)
Kroon, C.
1991 ‘Discourse Connectives and Discourse Type: the Case of Latin a t’. In:
Amsterdam in Budapest. Papers presented at the Vlth International Collo­
quium on Latin Linguistics by members of the Department of Latin of the
University of Amsterdam, 51-64
1995 Discourse Particles in Latin. A Study o f nam, enim, autem, vero and at.
Amsterdam: Gieben
Orlandini, A.
1994 ‘Si non eo die, at postridie (Cato agr. 1,2): une analyse pragmatique du
connecteur latin at’.Lalies 14,159-175
1995 ‘De la connexion: une analyse pragmatique des connecteurs latins atqui et
immo’. Lalies 15, 259-269
Ruijgh, CJ.
1971 Autour de τε épique. Études surla syntaxe grecque. Amsterdam: Hakkert
Wagner et Pinchon
1962 Grammaire dufrangais classique et moderne. Paris
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP

S.R . S l in g s
Vrijc Universiteit Amsterdam

1. In tro d u c tio n

1.1. P U SH an d P O P

To start with, let us consider the following fictitious dialogue:

(1) A: (a) So you wanted to write a thesis on particles, you told me.
B: ( b ) Yes, they’ve always intrigued me a lot.
A: (c) You don’t say. (d) Incidentally, I ’ve got a rather nice bottle of Aus­
tralian Chardonnay here, (e) y ’know, I don’t buy French stuff any more, I ’m
afraid, (f) Would you care for some?
B: ( g ) Yes, that would be very nice.
A: (Λ) Well, cheers! (/) Right, particles, huh? (j) D ’you realize how slippery
they are?

By means of clause (la), speaker A proposes a Discourse Topic, which is accepted

as such by B in clause (Id). However, before this Topic is worked out, there is what
in Discourse Analysis is known as an embedded sequence, starting at (Id) and end­
ing at (1h ) . At (li), A has to restate tire Discourse Topic, which is then further devel­
oped in ( I f ) . Embedded sequences have often been studied (the classic study is Jef­
ferson 1972), and it has been observed that very often their starting-point (also
known as PUSH, a term used first by Polanyi & Scha 1983) and the point where a
speaker returns to the embedding sequence (called POP) are marked by particles and
other relators. In this case, the PUSH is marked by i n c id e n ta lly in (Id) and the POP
by r i g h t in (li). In Greek, the most typical PUSH particle is γάρ, as n a m is in Latin
(Kroon 1995: 144-170), while the candidate for most typical POP particle is, of
course, οΰν (i g i t u r in Latin, cf. Kroon 1989). It has often been remarked that such
particles have the function of signposts: they help the interlocutor in his orientation
when processing the discourse.
Embedded sequences are characterised by the fact that they have a different
‘frame of reference’ from the embedding sequence. This explains frequent shifts in
tense and mood—contrast the past tenses of (1a b ) with the present tenses used from
(lc ) on—, shifts in deictic and anaphoric pronouns and so on. These are the
 S.R. SLINGS

PUSH/POP markers listed in Polanyi & Scha (1983). For continuous prose, I would
add shifts in type of text, from narrative to argumentative etcetera.
The analysis of spoken discourse in terms of hierarchical structure (embedded
vs. embedding) has been applied with great success to written discourse by the Swiss
scholar E. Roulet, in what has become known as the Geneva model.1 A crucial feature
in this approach is the consistent distinction made between central and subsidiary dis­
course units,12 and the possibility for the latter to be analysed in terms of centrality
and subsidiarity once again. Indeed all units which make up the embedded sequence
from the PUSH to the POP in (1) are not on the same level: there is embedding
within embedding. (Id) is a preparatory statement to the interactive move contained in
( I f ) , while (le) serves to explain the item ‘Australian’ in (Id). Schematically:

(la b e ) POP (10)

POP (1f g h )
PUSH (Id)
PUSH (le)

In this scheme a lower level indicates subsidiarity; thus, (le) is subsidiary to (Id),
(lde) together to ( I f ) and the whole sequence from (Id) through (1ft) to (li).
In this paper I will try to combine the Geneva model with the PUSH/POP model.
From the former, I borrow the analysis of text as hierarchical structures which allow
more than one degree of subsidiarity (in other words PUSH/POP sequences within
PUSH/POP sequences), from the latter the terminology and the possibility to identify
a PUSH/POP sequence without the obligation to indicate the direction of subsidiar­
ity—in other words, with regard to which specific clause or set of clauses outside a
subsidiary sequence the sequence is subsidiary.
In the study of Greek particles, very little attention has been paid so far to text
structure.3 The underlying assumption of most studies in this field is that text is a
monolithic sequence of clauses and sentences. The aim of this paper is to show that
this assumption leads to inadequate descriptions. Especially POP particles tend to
mark the coherence not between two successive clauses or sentences but between dis­
course units surrounding an embedded sequence, in other words coming before a
PUSH and after a POP. PUSH/POP analysis thus may contribute to a better unified
description of particles and other discourse markers.
I will try to show this for adversative particles, using as a corpus the work of
Herodotus, whose predilection for embedded sequences, at the levels of sentence,
paragraph and longer units is notorious—it may even be maintained that Book 2,
from chapter 2 onwards, is one long subsidiary stretch of text between a PUSH at

1 Cf. Roulet et al. (1985), Roulet (1991), Drescher & Kotschi (1988). Polanyi (1988) places
PUSH/POPs within a different theoretical paradigm.
2 1 use this term in order to avoid the distinction between acts and moves, which is compulsory in
the Geneva model, but which I have not found crucial in the study of my particular corpus. For the
same reason, I will not use the Geneva notion of ‘diaphony’.
3 The main exceptions are Bakker (1993) and Sicking & Van Ophuijsen (1993).
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 103

2.2.1 and a POP at 3.1.1. The particles studied from this perspective are άλλα,
μέντοι and καίτοι. δέ is not really an adversative particle, of course, and although it
is often used as a PUSH or POP particle, its basic function of marking textual seg­
ments (Bakker 1993) shows no difference in behaviour in hierarchical and non-hier-
archical sequences.
Here is a relatively simple example to illustrate how embedded sequences can be
identified in continuous written discourse:

(2) (a ) έσέβαλε μέν νυν στρατιήν καί οΰτος, έπείτε ηρξε, ές τε Μίλητον κ αί ές
Σμύρνην, καί Κολοφώνος το αστυ είλε. φ ) άλλ’ (c) ούδέν γάρ μέγα άπ'
αυτού άλλο έργον έγένετο βασιλεύσαντος δυών δέοντα τεσσεράκοντα
έτεα, φ ) τούτον μέν παρήσομεν τοσαΰτα έπιμνησθέντες. ’Άρδυος δέ
τού Γύγεω μετά Γύγην βασιλεύσαντος μνήμην ποιησομαι. ( d ) ούτος δέ
Πρνηνεας τε ε ίλε ές Μίλητόν τε έσέβαλε' (1.14.4-15.1)
‘ ( a ) Once established in power, Gyges sent a military expedition against
Miletus and Smyrna and captured Colophon, ( b ) However, (c) that being his
only act of any importance during a reign of thirty-eight years, φ ' ) we will
pass on without further comment and 111 mention his son Ardys who suc­
ceeded him. (d ) Ardys took Priene and attacked Miletus;'4

The embedding sequence consists of ( 2 a ) and ( 2 d )—note the similarity in verbs and
tenses (έσέβαλε, είλε) in both of these clauses. The embedded sequence (2b o b ') is
set off from it by a change in tense (the futures in (2b j ) and in person (first singular
and first plural). The first person plural also points to an intcractionality between
author and reader: in a conversation one might say ‘Well, we’ll skip him, won’t we.’
The point of the embedded sequence is, of course, to replace5 one Discourse Topic
(Gyges) with another (Ardys). There is a deictic shift after the POP: τούτον in ( 2 b ’ )
refers to Gyges, οΰτος in ( 2 d ) to Ardys. And finally the PUSH replaces a narrative
with auctorial comment, at the POP the narrative is resumed.
Again there is embedding within embedding: (2c) is one level further down than
( 2 b ) and (2b ’ ); it gives the justification for the Discourse Topic shift, plus some
additional information about the old Discourse Topic: the length of Gyges ’ reign. (2)
may therefore be rendered as follows:

(2a) POP (2 d )
PUSH (2 b ) POP (2b ‘ )
PUSH (2c)

It should be noted that POPs have a strong closing power: after ( 2 d ) it would be
impossible for the author to use Gyges as a Discourse Topic again (unless he were to
be formally reintroduced (cf. ex. (It)), as in fact he won’t be). Nor can Herodotus

4 Translation here and elsewhere Sélìncourt - Bum, slightly modified.

5 Cf. below on ‘replacing’ as the basic function of άλλά at PUSHes and (especially) POPs.
 S.R. SLINGS

use future and first-person verb forms any more, unless he opens another embedded
sequence.
Related to embedded sequences are what is normally referred to as ‘pre­
sequences’, which end with a POP without a previous PUSH. In Herodotus, these
logically occur at the beginning of a stretch of directly or indirectly reported speech.
Here is an example:

(3) (α) λάβε τον Μανδάνη έτεκε παιδα, φέρων δέ ές σεωυτοΰ άπόκτεινον
μετά δέ θάψον τρόπω οτεω αυτός βούλεαι. ό δέ αμείβεται' ( b ) Ώ
βασιλεύ, ούτε άλλοτε κω παρεΐδες άνδρί τφ δε άχαρι ούδέν,
φυλασσόμεθα δέ ές σέ και ές τον μετέπειτα χρόνον μηδέν έξαμαρτεΐν.
(c) άλλ' ε ϊ τοι φίλον τοΰτο οϋτω γίνεσθαι, ( d ) χρή δή τό γε έμόν
ΐητηρετέεσθαι έπιτηδέως. (1.108.4-5)
‘(α )... “Get hold of Mandane’s child—take it home and kill it. Then bury it
how you please.” Harpagus replied: “(b ) Sire, you have never had cause to
find fault with me yet, and I will take care not to offend you in the future, (c)
If this is your will, (d ) it is for me to do my duty and obey.”’

(3 b )is a pre-sequence, followed by a two-fold POP, one at the beginning of (3c),

which is marked by άλλά, another at ( 3 d ) , marked by δή. (3c) is a resuming Theme,
as εί clauses so often are; it has an anaphoric pronoun and pro-adverb (τοΰτο οΰτω),
which refer back to (3a); and of course (3 d ) is a response to the imperatives of (3a).
(3a) and ( 3 d ) are concerned with the future, ( 3 b ) with the past; it may be significant
that the first-person form in (3 b ) is replaced by τό γε έμόν in (3 d ) and that instead of
a first-person singular verb ‘I will do my duty,’ Harpagus uses an impersonal deon-
tic expression (χρή). It will be clear that pre-sequences and embedded sequences
cannot be fruitfully kept apart—it might even be maintained that Harpagus’ refusal to
respond directly to the king’s imperatives constitutes a PUSH in itself.
It is relevant to note that a POP may be realised in two stages, (3c) and ( 3 d )
respectively. This can be seen in (1) as well, both for the PUSH and the POP there:
(lc) seems to do little more than to prepare for the PUSH in (Id)—in other words,
(lc) is a temporary closure device for the embedding sequence, and likewise (1Λ)
prepares the POP, in which (10 may be said to be transitional as well.

1.2. d e re and d e d icto

In addition to the concepts of PUSH/POP analysis and of the Geneva model, I shall
operate throughout with the age-honoured distinction between use of relators d e re
and d e d ic to , as is also done by Daniele Torek in her recent study of causal relators in
spoken French (1996: 48).6 In the area of causal relators, this distinction coincides
with the familiar one between:

6 This is parallel to Schiffrin's distinction of internal and external causality (1985: 284). Cf.
Ruijgh (1971: 127): ‘causal proprement dit’ vs. ‘indirectemcm causal'. The distinction is extended
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 105

(4a) The streets are wet because it’s rained.

(4b) It must have rained because the streets are wet.

(4a) is usually called true cause or explanation { d e r e ) , (4b) justification { d e d ic io ) .

The difference is that in (4a) the b e c a u s e conjunct gives the cause of the fact stated
before it (‘A conjunct’), whereas in (4b) it gives the reason why the speaker states the
A conjunct.
The distinction has proved fruitful to me in the area of adversative relators as
well. Their use d e d ic to equals what is often identified as a subgroup of adversative
relations, called ‘rhetorical contrast’ by Kroon (1995: 213f.) (cf. Anscombre &
Ducrot 1977; Sweetser 1990: 101). I will try to show, however, that the use d e d ic io
represents a separate level of discourse. An adversative relation d e r e contrasts two
states of affairs, while a d e d i c t o relation opposes two propositions or speech-acts7
(cf. (5a) and (5c) below, as opposed to (5b) and (5d)). The possibility of accounting
for rhetorical contrast in this way, rather than as a subgroup of adversative relations,
was put forward tentatively by Kroon (1995: 216); PUSH/POP analysis of adversa­
tive relators in Herodotus strongly supports it.8
In other words, the distinction d e re: d e d ic to cuts across the familiar distinction
between replacing9 vs. denial of expectation.10 άλλά, which is typically a replacing
adversative, can be used both d e r e and d e d ic to ; and the same goes for μέντοι, which
is used for denial of expectation and for καίτοι, used for what I will call ‘inverted
denial of expectation’. Let me illustrate this by means of English examples for the
two types replacing and denial of expectation:

(5a) The dress is not red but green.

(5b) You look awful, but we’re not here to talk about your health. How’s your
work?
(5c) He is rich but unhappy.
(5d) You’re a nice enough chap, but stop talking all the time.

In (5a) one statement is replaced by another, in (5b) one Discourse Topic is dis­
missed and replaced by another. In (5c) an expectation raised by the statement ‘he is

to adversative panicles by Kroon (1995) (‘representational’ = de re vs. ‘rhetorical’ = de dicto). The

de dicto use is often called ‘epistemic’.
7 I do not wish to suggest that the distinction runs parallel to the various levels distinguished in
the Functional Grammar paradigm. Cf. Kroon ( 1996) on this issue.
8 Sanders el al. (1992) claim, using an entirely different approach from mine, that for every coordi­
nation relation a distinction can be made between semantic and pragmatic use (de re vs. de dicto);
thus also for denial of expectation and complex contrasts, which 1 shall not discuss—they say
nothing about replacing.
9 ‘Replacing’ or ’substitution’ are the terms most commonly used nowadays for what Denniston
(1954: Iff.) and Ruijgh (1971: 127ff.) call ‘eliminative’, also known as the sondern type. Cf.
Anscombre & Ducrot (1977); Spoorcn (1989: 9-11); Foolcn (1993: 112-114).
10 This term was first introduced in Lakoff's classic study of English but (1971). It corresponds
only partly to Denniston's ‘balancing’ (1954: 5ff.). See below on ‘modifying’, ‘inverted denial of
expectation’, ‘rebuttal’, and ‘rhetorical contrast’.
 S.R . S l i n g s

rich’ is denied—this analysis is impossible for (5d): it is not logical to assume that
since I think you’re a nice enough chap I may be expected to want you to talk all the
time. Rather, the fact that 1 say something positive about you may lead you to expect
that I am going to say something along the same lines—instead, I say something
negative.11
In the area of replacing adversatives, the distinction is borne out by the fact that
replacing adverbs (Eng. r a th e r , Greek τουναντίον; cf. Slings 1980: 118f.) can be
used as substitutes for replacing particles only when these are used d e re. Contrast:

(6a) The dress is not red but green.

(6b) The dress is not red. Rather, it is green.
(6c) You look awful, but we’re not here to talk about your health.
(6d) *You look awful. Rather, we’re not here to talk about your health.

For the PUSH/POP analysis of Herodotus, I prefer the d e r e : d e d ic to dichotomy to

Kroon’s more sophisticated tripartition, used with great success for three Latin b u t-
equivalents (1995: 58-62): representational (red), presentational ( a u te m ) and interac­
tional (a t) levels.1
112 In my material, all adversative relators are, at times, interactional.13
True, in more consistently dialogical texts, such an approach would be unsatisfactory.
But for entirely practical reasons, I will not use Kroon’s distinction in this paper.

Of the three most important ftw-equivalents, άλλα is virtually confined to POPs,

χαίτοι to PUSHes, while μέντοι is found with both. First, I shall therefore examine
άλλα and μέντοι as POP particles, then μέντοι and καίτοι as PUSH particles.14

2. ά λ λ α

2.1. d e re

On the discourse level, the use of άλλά in POPs mirrors its most frequent use within
the sentence, where it is virtually confined to replacing structures, of the type ‘Her
dress is not red but green.’ In fact, there may be a progressive ‘weakening’ in the
uses of άλλά: (i) prototypical clause-internal use as illustrated in (5a)—if a language
has a separate replacing particle (German s o n d e r n , Spanish s i n o ) it is used only in
this way; (ii) sentence- or clause-initial use d e r e , still preceded by a negative; (iii)

11 There is a contrast here between a Face Saving Act and a Face Threatening Act, cf. below on
(28).
12 This is a slight simplification (on my pari) as far as sed is concerned. For the tripartition, cf.
Halliday (1985: 36f.) on ‘the clause as a message’ (presentational), ‘the clause as an exchange'
(interactional), ‘the clause as a representation' (representational).
13 1 do not mean to say that all three arc interactional, but simply that Herodotus is too one-sided a
corpus to allow for an interactional status to be ascribed to any of the three.
14 άτάρ is used three times as a POP particle after a pre-sequence (3.29.2; 7.50.1; 8.144.1), never in
PUSHes. As the particle seems to be the equivalent of δέ in Herodotus, 1 will not discuss i t
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 107

sentence- or clause-initial use d e d ic to , where no preceding negative is necessary (cf.

(5b)).15 Of course, αλλά is not on a par with exclusively intra-clausal replacing par­
ticles like s o n d e r n or sin o : on the one hand, it is not the only particle that can be used
in the ‘not A but B’ type (δέ is a marginal competitor), on the other, even within the
clause or sentence its use is not restricted to replacing (for example, it can follow a
preparatory μέν). Still, given its intra-clausal use it is a fair assumption that αλλά is
prototypically a replacing adversative particle.
Normally in my material, a negative statement before the PUSH is replaced with a
positive one after the POP, the embedded sequence elaborating on the negative state­
ment Schematically:

(7) -i A PUSHelaboration of -> A rop αλλά B (A and B incompatible)16

An example:

(8) Ύπερβορέων δέ πέρι ανθρώπων ούτε τι Σκύθαι λέγουσι ούδέν ούτε

τινες άλλοι των ταύτη οίκημένων, PUSH εί μή άρα Ίσσηδόνες· ώς δ' έγώ
δοκέω, ούδ' οΰτοι λέγουσι ούδέν PUSH ελεγον γάρ αν και Σκύθαι, ώς
περί των μουνοφθάλμων λέγουσι. pop άλλ' Ήσιόδφ μέν έστι περί
Ύπερβορέων είρημένα, έστι δέ και Όμήρφ έν Έπιγόνοισι... (4.32)
O f the Hyperboreans we get no information from the Scythians or anyone
else in that part of the world, push except, perhaps, from the Issedones. I
don’t think the Issedones have anything to say on them either, push f°r >f
they did, we should have it from the Scythians too, like the story of the one-
eyed men. pop Those who do mention them are Hesiod, and Homer in the
E p i g o n i ...’ 17

The Discourse Topic (the Hyperboreans) is constant, one set of possible reporters on
them is replaced by another. It is in the nature of POPs that the Scythians and other
Russian peoples do not recur again with regard to the Hyperboreans. The embedded
sequence is set off from the embedding sequence by a clear difference in modality:
certain statements before the PUSH and after the POP, the ε ί μή άρα clause, the
δοκέω and the counterfactual construction in between.18

15 Some people may deny the term ‘replacing’ to this use, but I think the only way to account for
the difference between (5b) and (5d), and between de dicio use of ά λλο, μέντοι and καίτοι, is by
describing (5b) as replacing.
16 For incompatibility as a condition for replacing contexts, cf. Slings (1980: 116).
17 In Greek, ‘ μέν δέ is not a possible cluster, and άλλο ... μέν is the most frequently used alterna­
tive, cf. De Stryckcr & Slings (1994: 276f.) But I have not found that at PUSHes and POPs αλλά
... μέν behaves differently from άλλά without a following μέν.
18 Cf. 1.51.4; 88.3; 109.4 (T m not going to obey ... someone else must do the killing’); 2.15.3 (‘If
the lonians are right, the Egyptians wouldn't have had a country in the past... but they had one');
3.2.2 (‘the Egyptians are not mistaken ... they distort the truth'); 4.30.2 ( ‘mules cannot be bom in
Elis' ... description how they are bom abroad); 114.4 ('w e cannot live among your women ... let's
live apart’); 5.106.4 ( Ί did not plot this ... my deputy did it on his own’); 7.10Θ.1 (‘it must not
happen that the king takes part in the expedition (10η. 1)... he must stay at home’); 8.108.4 (‘we
 S.R . SLIN G S

Pre-sequences behave similarly: they normally give elaborations of an implied

(less often stated) -i A, the αλλά POP replaces this with B. As opposed to embedded
sequences of this type, pre-sequences are almost always about a rejected course of
action rather than a rejected statement. Schematically:

(9) elaboration of —■A ρορ άλλά B (A and B incompatible)

(3) is a case in point: (3b ) gives reasons for an unexpressed ‘I cannot disobey you’;
after the POP, which occurs at the beginning of (3c), this is replaced by Ί must obey
you.’ It is logical that very frequently there is a difference in speech act between the
pre-sequence and the main sequence; normally the POP marks a transition from
statement to directive. (3 d ) is atypical in that it contains not a directive but a statement
with the illocutionary value of a promise;*19 (14) below is more typical in this
respect.20
It is important to note that άλλά as a POP particle does not behave in the same
way with regard to pre-sequences as to (the majority of) embedded sequences. In the
cases symbolised by (7) άλλά replaces a statement or course of action mentioned
before the PUSH, and ignores the embedded sequence. As there is no PUSH in pre­
sequences, άλλά here replaces a course of action given immediately before the POP.
There is, besides, an interesting variant of the state of affairs schematised in (7),
in which άλλά likewise relates to the sequence directly preceding the POP. It con­
sists of a statement made prior to the PUSH recurring after the POP; the embedded
sequence elaborates on a contrary or contradictory statement. Schematically:

(10) A PUSH elaboration of -i (-> A) POP άλλά A

Here, of course, A and —.A are incompatible. For example:

(21) α ίτιο ςδε τ ο ύ τ ω ν έγένετο ό 'Ελλήνων θεός ε π ά ρ α ς έμέ στρατεύεσθαι.

push ούδείς γάρ οϋτω ανόητος έστι οστις πόλεμον προ ειρήνης αίρέε-
ταΐ' push έν μέν γάρ τη οί πάιδες τούς πατέρας θάπτουσι, έν δέ τφ οι
πατέρες τούς πάνδας. Pqp άλλά ταΰτα δαίμονί κου φίλον ήν οΰτω
γενέσθαι (1.87.3-4)
‘“The god of the Greeks encouraged m e to fight you: the blame is his. PUs h
No one is fool enough to choose war instead of peace — PUsh in peace sons
bury fathers, but in war fathers bury sons. POp It must have been heaven’s
will that this should happen.’”

must not destroy the bridges... wc must let the enemy escape’); 109.4 (‘we must not pursue the
enemy... it’s better to stay in Greece’).
19 The two arc of course related in that both promises and directives are action speech-acts.
20 Cf. 3.145.2; 4.115.2; 136.4; 5.79.2; 6.130.2; 7.5.2; 160.2; 8.2272; 5 7,2; 62.1; 143.1; 9 .9.2.
 ADVERSATIVE RELATORS BETW EEN PUSH AND POP 109

C r o e s u s ’s c l a i m t h a t i t w a s A p o l l o w h o c a u s e d h i m t o s t a r t th e w a r i s p r o v e d n e g a ­
t i v e l y i n t h e e m b e d d e d s e q u e n c e , th r o u g h e x c l u s i o n o f th e c o n t r a r y ( n a m e l y , t h a t h e
d i d s o o u t o f h i s o w n f r e e w ill) . A f t e r th e P O P t h e s t a t e m e n t is r e p e a t e d i n a m o d if ie d
f o r m . W h a t ά λ λ α r e p l a c e s h e r e is a n i m p l i e d n e g a t i v e s ta t e m e n t . T h e e m b e d d e d
s e q u e n c e i s a s e t o f g e n e r a l s ta te m e n ts i n th e p r e s e n t te n s e , c o n tr a s tin g w ith p a r tic u la r
s t a t e m e n t s i n th e p a s t te n s e . 21

2 .2 . d e d ic to

I n (3 ), (8 ) a n d (1 1 ), ά λ λ ά is u s e d de re. W h e n used de dicto α λ λ ά ro u n d s o ff ex cu r­

s u s e s a n d r e p l a c e s o n e D i s c o u r s e T o p i c w i t h a n o t h e r , th is is t h e o n l y ty p e o f c o n te x t
w h e r e t h e p a r t i c l e is f o u n d a t P U S H e s a s w e l l a s P O P s . ( 2 ) is a n e x a m p l e o f th e f o r ­
m e r . 22 W h e n ά λ λ α is s o u s e d a t P O P s , th e P O P is a l m o s t in v a r i a b l y r e a l i s e d i n tw o
s ta g e s :

(1 2 ) έ π ι σ τ ά μ ε ν ο ι ώ ν α υ τ ό τ ο ύ τ ο ο ΰ κ ά ν τ έ τ ε ν ν ο ν ά λ λ ’ ε ι κ ο ν , PUSH μ έ χ ρ ι δ σ ο υ
κά ρτα έδέοντο α υτώ ν, ώς δ ιέ δ ε ξ α ν ρυ5Η ώ ς γάρ δ ιω σ ά μ ε ν ο ι το ν
Π έ ρ σ η ν π ε ρ ί τ η ς έ κ ε ίν ο υ ή δ η τ ο ν α γ ώ ν α έ π ο ιε ΰ ν τ ο , π ρ ό φ α σ ιν τη ν
Π α υ σ α ν ίε ω ΰ β ρ ιν π ρ ο ϊσ χ ό μ ε ν ο ι ά π ε ίλ ο ν τ ο τη ν ή γ ε μ ο ν ίη ν τους
Λ α κ ε δ α ιμ ο ν ίο υ ς , p o p ά λ λ α τ α ΰ τ α μ ε ν ύ σ τ ε ρ ο ν έ γ έ ν ε τ ο , p op τ ό τ ε δ ε
ο ΰ τ ο ι ο ί κ α ί έ π ' Ά ρ τ ε μ ί σ ι ο ν Ε λ λ ή ν ω ν ά π ι κ ό μ ε ν ο ι ... ( 8 . 3 . 2 - 4 .1 )
‘I t w a s b e c a u s e th e A t h e n i a n s r e a l i s e d th is t h a t t h e y w a i v e d t h e i r c la im t o th e
com m and, push a n d th e y c o n t i n u e d t o d o s o a s l o n g a s G r e e c e d e s p e r a t e l y
n e e d e d t h e i r h e lp . T h i s w a s m a d e p l a i n e n o u g h b y t h e i r s u b s e q u e n t a c tio n ;
pu sh f ° r w h e n t h e P e r s i a n s h a d b e e n d r i v e n f r o m G r e e c e a n d th e w a r h a d
b e e n c a r r i e d to P e r s i a n t e r r i t o r y , t h e A t h e n i a n s m a d e t h e i n s u f f e r a b l e
b e h a v i o u r o f P a u s a n i a s t h e i r e x c u s e f o r d e p r i v i n g t h e L a c e d a e m o n i a n s o f th e
c o m m a n d . P 0P B u t th a t h a p p e n e d la te r; a t th a t m o m e n t w h e n th e G re e k s
a r r i v e d a t A r t e m i s i u m . ..’

T h e s t o r y - l i n e is i n t e r r u p t e d b e c a u s e H e r o d o t u s w a n t s t o i l l u s t r a t e t h e A t h e n i a n a tti­
tu d e w i t h r e g a r d to t h e c o m m a n d o f th e G r e e k f l e e t , ά λ λ ά n o t o n l y c l o s e s th e r e p o r t
o f th e la t e r e v e n t b u t a ls o d is m is s e s th e A th e n ia n s as D is c o u r s e T o p ic . T h e tw o
s t a g e s b y m e a n s o f w h i c h th e P O P i s r e a l i s e d a r e h e r e m a r k e d b y a μ έ ν a n d a δ έ
c la u s e re s p e c tiv e ly . T h e re is a n im p o rta n t d iffe re n c e b e tw e e n αλλά de re and de
dicto : as w e saw , a s a P O P p a rtic le u se d de re ά λ λ ά n o r m a l l y r e p l a c e s a s ta t e m e n t

21 Cf. 1.1 L 3 (the altern ativ es a re restated after th e em b ed d ed ώ ς ών eiause); 2 .4 3 .3 (43.2 ‘the
G reeks to o k H eracles over from th e E g y p tian s’ (em bedded sequence d ealing w ith the alternative)
‘H e ra c le s is an old E gy p tian g o d ’); 8.65.5; 9 .1 7 .4 ( ‘e v ery m an m ust sh o w th a t h e is a m an ’
(em bedded sequence ab out su n c n d c r) ‘let them know w h at it is to plo t m urder a g a in st G reeks’ (i.e.
fight fo r yo u r lives)).
22 Cf. 2.20.1; 7.19 0 adfin:, 231.
 S.R. Sl in g s

made before the PUSH. This is found with d e d ic to use as well, but more often, as
here, αλλά dismisses information given after the (second) PUSH.23
This type of replacing can also occur in speeches:

(13) Άνδρες "Ελληνες, λόγον έχοντες πλεονέκτην έτολμήσατε έμέ σύμμα­

χον έπί τόν βάρβαρον παρακαλέοντες έλθεΐν. push αύτ(Η δε έμεΰ
πρότερον δεηθέντος βαρβαρικοϋ στρατού συνεπάψασθαι... έπισκήπτον-
τός τε τον Δωριέος ... φόνον έκπρήξασθαι... ούτε έμεΰ ε'ίνεκα ήλθετε
βοηθήσοντες ούτε τον Δωριέος φόνον έκπρηξόμενοι, τό τε κατ' ύμέας
τάδε άπαντα ύπό βαρβάροισι νέμεται. pop άλλα push Υ^ρ ήμΐν και
έπί τό άμεινον κατέστη, pop νύν δέ push επειδή περιελήλυθε ό πόλεμος
και άπΐκται ές ύμέας, pop ούτω δή Γέλωνος μνήστις γέγονε (7.158.1-3)
‘“Greeks, do you have the face to come here and urge me with your selfish
arguments to help you resist a foreign invader? push Have you forgotten that
I, too, was once at war with a foreign power, and applied for your help? Yes,
and I begged you to avenge ... the murder of Dorieus... But what was your
answer? You refused to come either to help me or to avenge Dorieus’ death—
and for all you cared, this whole country might now be subject to foreign rule.
pop Well— push luck, as it happens, has come my way. pop Now PUsh since
the wheel has come full circle and it is you who are in danger of war—pop
you remember Gelon!”’

By means of the αλλά POP, Gelon dismisses his plight in the past as a Discourse
Topic—the following PUSH, marked by γάρ, explains why he is able to do so. The
POP after that (νύν δέ κτέ) returns to the opening of the speech (έμέ σύμμαχον έπί
τόν βάρβαρον παρακαλέοντες: Γέλωνος μνήστις γέγονε). This analysis may
help to explain the following difficult passage:

(14) λέγω δέ τοι ότι Ισον έστί πολλά τε και ολίγα λέγειν περί άποπλόου τού
ένθεΰτεν Πελοποννησίοισι. push έγώ γάρ αύτόπτης τοι λέγω γενό-
μενος ότι νύν ούδ' ήν θέλωσι Κορίνθιοί τε καί αυτός Εύρυβιάδης οΐοί τε
έσονται έκπλωσαι- push περιεχόμεθα γάρ ύπό τών πολεμίων κύκλω.
pop άλλ' έσελθών σφι ταΰτα σημηνον. (8.79.4)
‘“Let me tell you that the Peloponnesians may talk as much or as little as
they please about withdrawing from Salamis—it will not make the least
difference, push What 1tell you, I have seen with my own eyes: they c a n n o t
now get out of here, however much the Corinthians or Eurybiades himself
may wish to do so, push because our fleet is surrounded, pop So go in and
tell them that!’”

23 Cf. 1.75.6; 4.77.2 (‘Anacharsis was killed by the Scythian king’, different version, ‘this story,
however, is only a frivolous Greek invention; the plain truth is that Anacharsis was killed in the
way 1 have described’); 7.171.1; 9.71.4 (cf. (41) below). Dismissing information given before the
PUSH: 4.16.2 ('Nobody has any accurate information about this region ... 1 will write down what
careful inquiry has told me’); 8.119 ad jin:, 9.85.1.
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 111

Aristides tells Themistocles that withdrawal is no longer an option: it may perhaps be

maintained that άλλά indicates that withdrawal must be replaced as a Discourse
Topic (in the deliberation of the Greek commanders) by the surrounding of Salamis.
The embedded sequence gives the reasons for this. Note the thematic links between
the sequence preceding the first PUSH and the first part of the embedded sequence
(λέγω δέ τοι : τοι λέγω; περ'ι άποπλόου : έκπλωσαι). The anaphoric ταΰτα after
the POP can no longer refer to the withdrawal but must refer to the message given
after the second PUSH. Again, the POP marks a speech-act transition.
Once, άλλά is used d e d i c t o after a pre-sequence:

(15) TH κε μέγ' οίμώξειε ό Πελοπίδης 'Αγαμέμνων πυθόμενος Σπαρτιήτας

την ήγεμονίην άπαραιρήσθαι ΰπό Γέλωνός τε καί Συρηκοσίων. ΡΟρ
άλλά τούτου μεν τοΰ λόγου μηκέτι μνησθης... (7.159)
‘“Agamemnon, son of Pelops, would groan in his grave if he heard that
Sparta had been robbed of her command by Gelon and his Syracusans! pop
Let us hear no more of this ...”’

A Discourse Topic is rejected explicitly.

2.3. Denial of expectation

In all instances discussed so far, άλλά is clearly recognisable as the replacing con­
nective. There are, however, some examples, four in all, where it is used at POPs for
denial of expectation, the other major class of adversative contexts. Remarkably, the
contexts of these instances are closely similar semantically: in all of them an effort to
achieve something is described before the PUSH, failure to achieve it after the POP:

(16) Ξέρξης δέ παντοιος έγίνετο ού βουλόμενος δούναι, push κατ’ άλλο μεν
ούδέν, φοβεόμενος δέ "Αμηστριν, μή καί πριν κατεικαζούση τά γινό­
μενα οϋτω έπευρεθη πρήσσων pop άλλά πόλις τε έδίδου καί χρυσόν
άπλετον καί στρατόν, τοΰ έμελλε ούδείς όρξειν άλλ' ή εκείνη· push
Περσικόν δέ κάρτα ό στρατός δώρον. pQp άλλ' push ού γάρ έπειθε, pop
διδοΐ τό φάρος· (9.109.3)
‘Xerxes did everything he could think of to get out of giving it, push simply
because he was afraid that Amestris had already guessed what was going on
and would now find her suspicions confirmed. pqP He offered her cities, gold
in unlimited quantity, an army under her sole command—push a thoroughly
Persian gift—pop but PUsh he couldn’t get her off the robe, pop so he gave it
to her.’

The first άλλά at the first POP is simply replacing: he didn’t want to give the woman
the robe but offered her cities and other luxuries. But the άλλά at the second POP
cannot be so explained: it contrasts Xerxes’ attempt to get out from under his
promise with his failure to do so—there is a strong thematic continuity before the
 S.R. SLINGS

first PUSH (ού βουλόμένος δούναι), after the first POP (έδίδου) and after the sec­
ond (διδοι)·24 There is one similar case in a pre-sequence:

(17) σύ δή κείνος εις ό άποσπεύδων Ξέρξην στρατεύεσθαι èrti τήν Ελλάδα

ώς δή κηδόμενος αύτοΰ; pop άλλ' ούτε ές τό μετέπειτα ούτε ές τό
παραυτίκα νύν καταπροΐξεαι άποτρέπων τό χρεόν γενέοθαι' (7.17.2)
“‘Are you the man who in would-be concern for the king is trying to dis­
suade him from making war on Greece? pgp You will not escape unpunished,
either now or hereafter, for seeking to turn aside the course of destiny;”'

Artabanus’ attempt to dissuade Xerxes is contrasted with his failure as predicted by

the dream. It is interesting to note that a mirrored version of this type of context
(failure contrasted with effort) is found with καίτοι PUSHes, cf. (31).
These examples, to which two άλλά γάρ POPs should be added (cf. below) are
the only ones in which άλλά does not function as a replacing adversative, whether at
PUSHes or at POPs. My explanation is that we observe here a gradual encroaching
of άλλά on the denial-of-expectation area, in other words, of άλλά trying to ‘replace’
μέντοι, in the type of context which is perhaps most prototypically one in which an
expectation is denied.25

2.4. Conclusion

With the exception of this small group, άλλά is always used for replacing as a
PUSH or POP panicle. This is in marked contrast to its use (opening sentences) in
non-hierarchical sequences, where it is frequently found in denials of expectation.
This illustrates the imponance of PUSH/POP analysis. Apparently, PUSHes and
POPs are more similar to intra-clausal contexts than parallel sequences are.

2.5. άλλά γάρ

I round off my discussion of άλλά with an analysis of άλλά γάρ.26 Given the fact
that άλλά is used overwhelmingly in POPs, whereas, as I said, γάρ is the most nor­
mal PUSH particle, this is a curious cluster. In fact it is not very frequent in
Herodotus: it is found about a dozen times. The cluster always announces a future
POP; it is itself used in POPs five times. Here is an example of the simple ‘not A but
B' type:

(18) οΰτος ώνήρ ούκ έφη παν όρθως τούς χρηομολόγους συμβάλλεσθαι, push
λέγων τοιάδε, ε ί ές 'Αθηναίους ε ίχε τό έπος είρημένον έόντως, ούκ αν

24 Cf. 3.36.3; 152 (άλλ' ούδ' ώς).

25 Cf. Kroon (1995: 215), who appears to use an ‘effort and failure’ example as the most representa­
tive of denial of expectation.
26 I do not discuss άλλά... γάρ, except for άλλ’ ού γάρ, as άλλά γάρ ού does not occur. But I dis­
regard cases where ούγάρ opens a parenthesis as in (16).
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 113

οϋτω μιν δοκέειν ήπίως χρησθήναι, άλλα ώδε *Ω σχετλίη Σαλαμίς, άντί
του ΤΩ θείη Σαλαμίς, push ε ΐ πεΡ Υε έμελλον οί οίκήτορες άμφ' αυτή
τελευτήσειν. ΡΟΡ αλλά γάρ ές τούς πολεμίους τω θεώ είρήσθαν τό
χρηστήριον συλλαμβάνοντι κατά το ορθόν, άλλ' ούκ ές 'Αθηναίους, pop
παρασκευάζεσθαι ών αυτούς ώς ναυμαχήσοντας συνεβούλευε ...
(7.143.1-2)
‘This man declared that there was an important point in which the interpreters
were mistaken. push If- he maintained, the disaster referred to was to strike
the Athenians, it would not have been expressed in such mild language.
“Hateful Salamis” would surely have been a more likely phrase than ‘divine
Salamis,’ push *f the inhabitants of the country were doomed to destruction
there, pop On the contrary, the true interpretation was that the oracle referred
not to the Athenians but to their enemies, pop So he advised them to prepare
at once to meet the invader at sea...’

At the first POP the false interpretation of the oracle is replaced with the true one, but
this in itself is presented as subsidiary to Themistocles’ actual proposal, which comes
at the second POP; the subsidiarity is marked by means of ών at the second POP.
The following is an example of άλλα γάρ used d e d i c t o :

(19) έπεί ώς γέ τι μάλλον οΰτοι "Ιωνές είσι των άλλων Ίώνων ή κάλλιόν τι
γεγόνασι, μωρίη πολλή λέγειν PUsh (arguments for this view: these loni-
ans are in fact of very mixed descent) p^p αλλά γάρ περιέχονται του ούνό-
ματος μάλλον τι των άλλων Ίώνων ΡΟΡ εστωσαν δή καί οί καθαρώς
γεγονότες Ίωνες. (1.146.1-147.1)
‘It is quite absurd to pretend that the inhabitants of the twelve cities of the
Ionian league are any more Ionian, or of purer blood, than the Ionians gen­
erally p u sh ··· pop ®ut since these Ionians set more store by the name than
the others, pop certainly the Ionians of pure blood must be allowed to be
called Ionians as well.’27

The άλλα γάρ POP dismisses the lengthy discussion of the mixed origin of the
Ionians in Asia. The fact that these are more attached to the name than others is sub­
sidiary to the ironical remark that follows at the second POP.
Here too, one finds denial of expectation, again in effort and failure contrasts:

(20) push καίτοι πάντα σοφίσματα καί πάσας μηχανάς έπεποιήκεε ές

αυτούς Δαρείος' Ρ0Ρ άλλ' ούδ' ώς έδύνατο έλειν σφεας, push άλλοισί τε
σοφίσμασι πειρησάμενος καί δή καί τω Κύρος ειλέ σφεας, καί τούτω
έπειρήθη' ροΡ άλλα γάρ δεινώς ήσαν έν φυλακήσι οί Βαβυλώνιοι, Ρ0Ρ
ούδέ σφεας οΐός τε ήν έλεΐν. (3.152)

27 The translation of the last clause, whose interpretation is disputed, follows Van Groningen
(1959: 73). Cf. 9.27.4.
 S.R . SLIN G S

‘push Every trick, every possible device, had been tried; pop but to no pur­
pose. The town could not be taken, push not even when Darius, after all else
had failed, attempted to repeat the method which Cyrus had previously used
with success. P0P The Babylonians were always on the watch with extraordi­
nary vigilance, pop and he could not take the city.’

The situation is different from the one analyzed under (16) in that the embedded
sequences describe the effort, the POPs the failure. The άλλα γάρ POP gives the
cause of the failure.28 The passage also illustrates that subsidiarity or embedding
from a discourse point of view need not go hand in hand with syntactic subordina­
tion: the two last clauses are coordinated by means of ούδέ, yet the first is subsidiary
to the second (cf. (13), where a δέ clause follows upon άλλα ... γάρ).

3. μ έντοι

Since μέντοι is etymologically related to non-preparatory μέν (Attic μήν), I will start
the treatment of μέντοι with a remark on the relationship between μέντοι and
μέν/μήν. At first sight, it might look as if the two are, in Herodotus at any rate, in
complementiiry distribution: μέντοι is hardly ever used as a modal particle, μέν fre­
quently.29*As a connective particle, μέν is confined to the clusters καί μέν and ού
μέν ουδέ: there is no καί μέντοι and no ού μέντοι ούδέ in Herodotus 20 although
these clusters are found in Attic (cf. Denniston 1954: 414; 410).31 But appearances
can be deceptive: there is little or no similarity between μέν and μέντοι, whether at
PUSH/POPs or elsewhere. The only μέν cluster to occur at PUSH/POPs with any
frequency at all, ού μέν ούδέ, is hardly ever found in contexts which can be usefully
called adversative, so I leave the particle out of account altogether.
Within the clause, μέντοι is entirely different from άλλα. Unlike άλλα, μέντοι
is not used for replacing false statements with true ones, or wrong courses of action
with right ones: in A μέντοι B both A and B are true statements (far less frequently
right courses of action). Normally, B is a denial of an expectation raised by A, or an
expectation raised by B can be a denial of an expectation raised by A,32 or B may be
a modification of A,33 but what matters most is the imbalance between them: the
speaker attaches more value to B than to A.

28 C f. 9.113.2.
29 μ έν το ι in the p h rase κ α ί τα ΰ τα μ έν το ι; μ έ ν in the clusters μή μ έ ν a n d η μέν.
313 ού μ έν το ι ο ύ δ έ is a va ria n t a t 6.45.1. ο ύ μ έ ν ο ύ δ έ o r ού μήν ο ύ δ έ is unanim ously transm itted
ten lim es (editors change μήν to μ έν ).
31 ού μ έν το ι ο ύ δ έ is very rare in A ttic, b u t ο ύ μήν ο ύ δ έ q u ite c o m m o n , cf. D en n isto n (1954:
338).— O nly in γ ε μ έ ν and γ ε μ έ ν τ ο ι is th e re a nything lik e an o v e rla p , b u t the fig u res a re not
im pressive: four tim es γ ε μ έν (once in an oracle), once γ ε μ έν το ι. C f. Po w e ll (1938) r .w .
32 Cf. K roon (1995: 211f.) o n direct vs. in d ire ct concession. I m a d e th e sam e distinction (1980:
122).
33 T his is a subclass o f adversative relatio n s discu ssed in B lo m q v ist (1969: 21) (exam ple: κ α λ ό ς
έ σ τιν , ού μήν τ α όμ μ α τά γ ε ) b u t not o ften reco g n ised in d iscu ssio n s o f adversativity. I t differs
from denial o f e xpectation in that th ere is n o association o f A co ntradicted b y B . W h at m odification
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 115

3.1. POP particle, d e re

After embedded sequences, μέντοι usually opposes a B statement to an A state­

ment—more precisely: denies an expectation raised by an A statement, or modifies an
A statement—made before the PUSH. Schematically:

(21) A push elaboration of A p0p μέντοι B (B denial exp. A)

I give two examples, one of denial of expectation proper, one of modification:

(22) άντιπέμπει προς ταΰτα ή Φαιδυμίη φαμένη κινδυνεΰσειν μεγάλως, ήν

ποιη ταΰτα' push 6*· γάρ δή μή τυγχάνει τά ώτα εχων, έπίλαμπτος δέ
άφάσσουσα εσται, εΰ είδέναι ώς άίστώσει μ ιν pop όμως μέντοι ποιή-
σειν ταΰτα. (3.69.4)
‘Phaedymia answered that it would be an extremely risky thing to do; push
for if her husband proved to have no ears, and she were caught feeling for
them, he would be certain to kill her. pop Nevertheless she was willing to take
the risk.’

If something is risky, one might expect that she will not do it, but the expectation is
contradicted.*
34

(23) καλέσας τόν βουκόλον τόνδε παραδίδωμι τό παιδίον, φας σε γε είναι

τόν κελεύοντα άποκτεΐναι αυτό' push κ αι λέγων τοΰτό γε ούκ έψευ-
δόμην push ένετέλλεο οϋτω. Pqp παραδίδωμι μέντοι τωδε κατά
τάδε, έντειλάμενος θεΐναί μιν ές έρημον όρος ... (1.117.4)
“‘... Γ gave the child to him, and told him it was your orders that it should be
killed. PUsh And that was the truth: PUsh was indeed what you commanded.
pop I gave it to him, however, with careful instructions, telling him to expose it
in a lonely place amongst the hills

The statement παραδίδωμι τό παιδίον is repeated after the POP, with an important
modification.35

and denial of expectation have in co m m o n , however, is the fa c t that th ey can both be paraphrased by
. means of a sentence of the 'although A, B' type.
34 The collocation όμως μέντοι is curious because the two words seem to do the same job. όμως is
a pro-adverb denoting denial of expectation anaphorically (only rarely cataphorically), for example
after an εί καί clause. In άλλ' δμως (7.148.4; 8.143.1) άλλα is a replacing adversative used de
dicto, so there is no tautology. Perhaps όμως is a Theme-like constituent here, as English still often
is. Cf. 1.189.4.
35 An alternative analysis would be denial of expectation d e dicto: from Harpagus’ admission that
he had not killed the baby himself but given it to the shepherd the king may have inferred that
Harpagus had therefore been disobedient, and this is denied in the following. For further examples
of μέντοι de re used as symbolised in (21) cf. 7.233.2; 9.40 (modifying); 111.1.
 S .R . SL IN G S

3.2. P O P particle, d e d ic to

μ έ ν τ ο ι is n o t u s e d de re a f t e r p r e - s e q u e n c e s ; m o r e i n g e n e r a l , i t s tr o n g l y f a v o u r s de
dicto u s e a s a P U S H / P O P p a r tic le . A t t h e de dicto l e v e l, th e d i f f e r e n c e b e tw e e n ά λ λ ά
a n d μ έ ν τ ο ι c o m e s o u t as c le a r ly a s a t t h e de re le v e l. A s w e s a w i n e x a m p le s (2 ) , (1 2 )
a n d (1 4 ) , ά λ λ ά r e p l a c e s o n e D i s c o u r s e T o p i c a n d /o r s t o r y - l i n e w i t h a n o th e r , μ έ ν τ ο ι ,
o n t h e o t h e r h a n d , i s u s e d t o r e d i r e c t t h e r e a d e r ’s e x p e c t a t i o n 36 r a i s e d b y t h e e m b e d ­
d e d seq u en ce:

(2 4 ) o 'i έ σ έ β α λ ο ν μ έ ν έ ς τ η ν Ά σ ί η ν Κ ι μ μ ε ρ ί ο υ ς έ κ β α λ ό ν τ ε ς ε κ τ η ς Ε υ ρ ώ π η ς ,
τ ο ύ τ ο ισ ι δέ έ π ισ π ό μ ε ν ο ι φ ε ύ γ ο υ σ ι ο ΰ τω ές τη ν Μ η δ ικ ή ν χώ ρην
ά π ίκ ο ν τ ο . push έσ τι δέ άπό τ η ς λ ίμ ν η ς τ η ς Μ α ιή τ ιδ ο ς έ π ί Φ ά σ ιν
π ο τ α μ ό ν κ α ί έ ς Κ ό λ χ ο υ ς τ ρ ιή κ ο ν τ α ή μ ε ρ έ ω ν ε ύ ζ ώ ν ω ο δ ό ς, έ κ δ έ τη ς
Κ ο λ χ ίδ ο ς ο ύ π ο λ λ ό ν ύ π ε ρ β ή ν α ι έ ς τ η ν Μ η δ ικ ή ν , ά λ λ έ ν τό δ ιά μ έ σ ο υ
έ θ ν ο ς α υ τ ώ ν έ σ τ ι, Σ ά σ π ε ιρ ε ς , τ ο ύ τ ο δ έ π α ρ α μ ε ιβ ο μ έ ν ο ισ ι ε ί ν α ι έ ν τη
Μ η δ ικ ή , pop ού μ έντοι ο ι γ ε Σ κ ύθα ι τα ύτη έσέβαλον, ά λλα την
κ α τ ύ π ε ρ θ ε ο δ ό ν π ο λ λ φ μ α κ ρ ο τ έ ρ η ν έ κ τ ρ α π ό μ ε ν ο ι, έ ν δ ε ξ ιή έ χ ο ν τ ε ς τό
Κ α υ κ ά σ ι ο ν ό ρ ο ς . ( 1 .1 0 3 . 3 - 1 0 4 .2 )
‘T h e S c y t h i a n s h a d e n t e r e d A s i a i n p u r s u i t o f t h e C i m m e r i a n s w h o m th e y
h a d e x p e l l e d f r o m E u r o p e , a n d s o th e y h a d i n v a d e d M e d i a n t e r r i t o r y , push

F r o m t h e s e a o f A z o v to t h e P h a s i s a n d th e C o l c h i a n s i s a t h i r t y d a y s ’ j o u r ­
n e y f o r a q u i c k tr a v e lle r ; b u t i t i s n o t f a r f r o m C o l c h i s t o M e d ia , a n d to r e a c h
i t o n e h a s t o p a s s th r o u g h t h e c o u n t r y o f o n l y o n e i n t e r v e n i n g p e o p l e , t h e
S a s p ir e s . P 0 P T h e S c y t h i a n s , h o w e v e r , d id n o t e n t e r M e d i a b y th is w a y , b u t
t o o k th e m u c h l o n g e r n o r th e r n r o u t e , k e e p in g th e C a u c a s u s m o u n ta in s t o th e ir
r ig h t.’

N o t e th e r e p e t i t i o n o f έ σ έ β α λ ο ν b e f o r e t h e P U S H a n d a f t e r th e P O P , a n d t h e c o n ­
t r a s t b e tw e e n th e a o r is ts a n d th e o m n i t e m p o r a l p r e s e n t s o f t h e e m b e d d e d s e q u e n c e , a
g e o g r a p h ic a l e x c u r s u s w h ic h m a y e a s i l y l e a d th e r e a d e r t o e x p e c t th a t t h e S c y t h i a n s
h a d t a k e n th e r o a d a c r o s s t h e l a n d o f t h e S a s p ir e s ; μ έ ν τ ο ι is u s e d t o c o n t r a d i c t th is
e x p e c ta tio n . A s i n th e c a s e o f ά λ λ ά , μ έ ν τ ο ι de dicto m a y m o d ify th e e m b e d d e d
r a t h e r th a n t h e e m b e d d i n g s e q u e n c e . 37 I n s p e e c h e s , μ έ ν τ ο ι c a n r e d i r e c t th e l i s t e n e r ’s
e x p e c ta tio n t h a t d i e s p e a k e r is g o in g to s a y s o m e th in g w h ic h h e a c tu a lly w ill n o t s a y :

(2 5 ) (th e M y s ia n s a sk C ro e s u s f o r h is s o n a n d a s e le c t b a n d o f y o u n g m e n )
π α ιδ ό ς μ έ ν π έ ρ ι το ύ έ μ ο ΰ μ ή μ ν η σ θ ή τ ε έ τ ι· pu sh ού γάρ α ν ΰ μ ιν

36 Cf. Sicking (1993: 34) ‘correcting p o ssib le misconceptions an d /o r frustrating expectatio n s’.
37 Cf. 4.81.2 (from th e fact that H erodotus says th at he w as un ab le to reach a conclusion a b o u t the
num ber o f th e S cythians a reader m ight have in ferred th at h e will p a ss o v er the problem , b u t h e goes
on to tell abo u t th e E xam paeus bow l, w h ich g iv es an in dication— referen c e to th e e m b ed d in g
sequence); 9.15.3 (the position occupied b y the Persian army as describ ed in the em bedded sequence
m ight lead to the assum ption th at the palisade M ardonius con stru ed co v ered all th at g round, b u t it
did not).
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 117

σ υ μ π έ μ ψ α ιμ ν push ν ε ό γ α μ ό ς τ ε γ ά ρ έ σ τ ι κ α ί τ α ΰ τά ο ί ν ΰ ν μ έ λ ε ι, pop
Λ υ δ ώ ν μ έ ν τ ο ι λ ο γ ά δ α ς κ α ί τ ό κ υ ν η γ έ σ ι ο ν π α ν σ υ μ π έ μ ψ ω ... ( 1 .3 6 . 3 )
“ Ί f o r b i d y o u to m e n t i o n m y s o n . pu sh I c o u ld n o t s e n d h im ; pu sh h e is
j u s t m a rrie d , a n d th a t k e e p s h im b u sy . P0P B u t I w ill c e r ta in ly s e n d p ic k e d
m e n , w i t h a c o m p l e t e h u n t i n g o u t f i t . ..’”

G iv e n th e f a c t th a t C ro e s u s s a id n o to th e f ir s t p a r t o f th e ir r e q u e s t, th e M y s ia n s
m i g h t h a v e e x p e c t e d t h a t h e is a l s o g o in g t o s a y n o t o t h e s e c o n d p a r t ; th i s e x p e c t a ­
tio n is d e n ie d . A s o p p o s e d to (2 4 ), μ έ ν τ ο ι h e re d e n ie s a n e x p e c ta tio n r a is e d b e fo re
th e P U S H .38 T h e r e is o n e e x a m p l e f r o m a c o n v e r s a tio n :

(2 6 ) ... τ ρ α π έ ζ η π α ρ α κ α τ η μ έ ν ω ν λ α β ο ΰ σ α ν θ ρ ί δ α κ α τ η ν γ υ ν α ί κ α π ε ρ ι τ ί λ α ι
κ α ί έ π α ν ε ιρ έ σ θ α ι τ ο ν α ν δ ρ α κ ό τ ε ρ ο ν π ε ρ ιτ ε τ ι λ μ έ ν η ή δ α σ έ α ή θ ρ ίδ α ξ
έ ο ΰ σ α ε ί η κ α λ λ ιώ ν , κ α ί τ ό ν φ ά ν α ι δ α σ έ α ν , τη ν δ έ ε ί π ε ΐ ν p o p Τ α ύ τ η ν
μ έ ν τ ο ι κ ο τ έ σ υ τ η ν θ ρ ίδ α κ α έ μ ιμ ή σ α ο , τ ο ν Κ υ ρ ο υ ο ΐκ ο ν ά π ο ψ ιλ ώ σ α ς .
( 3 .3 2 .3 - 4 )
‘ ... t h e t w o w e r e s i t t i n g a t t a b le , w h e n t h e w o m a n to o k a l e t t u c e a n d , a f t e r
p u l l i n g o f f th e le a v e s , a s k e d h e r h u s b a n d w h e t h e r h e th o u g h t i t l o o k e d b e tte r
w i t h i t s le a v e s o f f o r o n . C a m b y s e s s a i d t h a t h e p r e f e r r e d i t b e f o r e i t w a s
s tr ip p e d , w h e r e u p o n th e w o m a n r e p lie d : p o p “ Y o u h a v e d o n e s o m e th in g
s i m i l a r t o w h a t I d i d t o t h e le ttu c e : y o u h a v e s t r i p p e d th e h o u s e o f C y r u s
b a r e . ’”

I ta k e t h e f i r s t p a r t o f th e c o n v e r s a t i o n a s a p r e - s e q u e n c e , μ έ ν τ ο ι p r o b a b l y i n d i c a t e s
th a t th e c o n v e r s a tio n w a s n o t in f a c t a b o u t v e g e ta b le s , as C a m b y s e s m a y h a v e b e e n
l e d to b e l i e v e , b u t a b o u t th e m u r d e r o f S m e r d is . I f th i s a n a l y s i s is c o r r e c t , μ έ ν τ ο ι is
s till u s e d f o r d e n ia l o f e x p e c ta tio n , e v e n th o u g h it c a n n o lo n g e r b e tr a n s la te d b y
m e a n s o f a n E n g l i s h a d v e r s a t i v e r e l a t o r . 39 A s t h e q u e e n d i s m i s s e s v e g e t a b l e s a s a
D i s c o u r s e T o p i c , th is, p a s s a g e c o m e s c l o s e t o t h e o n e s w h e r e α λ λ ά is u s e d de dicto,
c f . e s p e c i a l l y ( 1 8 ) . 4 0 ·, -
I n p r e - s e q u e n c e s ’, μ έ ν τ ο ι n o r m a l l y d e n i e s th e e x p e c t a t i o n , r a i s e d b y w h a t a p e r ­
s o n s a y s i n th e p r e - s e q u c n c e , t h a t h e w ill s a y A — h e s a y s B in s te a d . S c h e m a tic a lly :

(2 7 ) A p o p μ έ ν τ ο ι B (B d e n i a l e x p . A )

38 C f. 4.163.3 (σύ μ έν το ι); 6 .8 6 α .1 (referen ce to se q u en c e b e fo re the PU S H ); 7 .1 0 1 .3 (= (29)

below ); 8.29.2; 142.3 (after h a v in g b lam ed the w ar o n the A thenians, the Sp artan s unexpectedly
express com passion with their p light— reference to sequence befo re the PUSH).
39 P redictably, D enniston classifies th is passage under ‘p ro g re ssiv e ' (1 9 5 4 :4 0 9 ). It needs hardly to
be said that a separate ‘progressive’ valu e o f μ έν το ι is highly doubtful: it is the o u tco m e o f his use
o f translation as principal criterio n fo r classification, w hich w as p erhaps inevitable in h is day, but
w hich m akes m ost o f his analyses unacceptable.
40 A n altern a tiv e in terp retatio n , su g g ested to m e in d e p en d e n tly b y C aro lin e K ro o n an d R odie
R isselada, is to m ake μ έν το ι rea ct to C am byses’ answ er (‘i f you think the lettuce is m o re beautiful
w ith its leaves on, I w ould have e x p ected you to a c t sim ilarly w ith regard to the h o u se o f C y ru s’).
I f so , this is no t a p re-sequence and μ έ ν τ ο ι not a P O P -p article (o f course, it is still u sed fo r denial
o f expectation de diclo).
 S.R . SLIN G S

This is often found in politeness strategies. A potential face-threatening act (FTA)

after the POP is neutralised by one or more face-saving acts (FS As) before it:

(28) ΤΠ παΐ Ύστάσπεος, εις τε πατρός άγαθοΰ καί έκφαίνειν οΐκας σεωυτόν
έόντα του πατρός οΰδέν ήσσω' pop την μέντοι έπιχείρησιν ταύτην μή
οϋτω συντάχυνε άβούλως, άλλ' έπ ί το σωφρονέστερον αύτην λάμβανε'
(3.71.3)
‘“You are the son of a brave father, Darius, and seem likely to prove as good
a man as he; POp nevertheless I advise you not to be rash or in too much of a
hurry. What we need is prudence.’”

The compliments paid in the pre-sequence are followed up by a face-threatening

appeal to give up the plan.41 Once, we find an FTA in an embedded sequence:

(29) νυν <δν μοτ τάδε φράσον, e i "Ελληνες ύπομενέουσι χ είρ α ς έμοί
άνταειρόμενοι. push Υάρ, ώς έγώ δοκέω, ούδ' ε ΐ πάντες Έλληνες
καί οί λοιποί οι προς έσπέρης οίκέοντες άνθρωποι συλλεχθείησαν, ούκ
αξιόμαχοι είσ ι έμέ έπιόντα ύπομειναι, μή έόντες άρθμιοι. pop έθέλω
μέντοι καί το άπό σεΰ, όκοΐόν τι περί αυτών λέγεις, πυθέσΘαι. (7.101.2-
3)
‘“Tell me, then—will the Greeks dare to lift a hand against me? PUSH My
own belief is that all the Greeks and all the other western peoples gathered
together would be insufficient to withstand the attack of my army, if they are
not united, pop But it is your opinion on the subject that I should like to hear
as well.’”

Xerxes asks Demaratus’ opinion, but he gives his own first, which is face-threaten­
ing. The neutralisation of the FTA works only because of the identity of the directives
before the PUSH and after the POP.42 It stand to reason that αλλά is never used in
politeness strategies. As a replacing adversative, άλλά would deny the validity or rel­
evance of a face-saving statement and so frustrate the politeness strategy, μέντοι, on
the other hand, as a denial-of-expectation adversative, is the proper particle to balance
an FTA and an FSA.43

41 Cf. 1.155.3; 7.13.2. R elated (p rologue to giving an o p in io n , w h ich is p o ten tially an FT A ):

5.111.3; 8.102.1. In so m e o f these passages ed ito rs p rin t a com m a be fo re the μ έν το ι con ju n ct, b u t I
have taken them in to consideration because th e preceding conjunct is subsidiary. N o t in a politeness
strategy: 7.160.1.
42 Cf. 1.120.4 (pre-seq u en ce). T w o F T A s surro u n d in g an e m b ed d e d sequence (th ere fo re n o t a
politeness strategy): 8.29.2.
43 In (3) ά λ λ ά repla ce s an explicitly n egated F T A w ith an FS A , so th ere is n o politeness strategy.
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 119

3.3. P U SH particle, d e re

As a PUSH particle, μέντοι is typically used to open small excursuses, often found
at the end of a story or, rarely, description. The excursus always belongs to the same
type of text as the narrative main sequence, and there is almost always Discourse-
Topic continuity before and after the PUSH—as we will see, the same holds good for
καίτοι PUSHes; POPs, whether marked by άλλα or by μέντοι may or may not have
such continuity. Schematically:

(30) A puSH μέντοι B (B denial exp. A)

Once again, the use of μέντοι d e r e is in a minority—it may even be non-existent.

The following instance admits of a d e r e interpretation:

(31) Μήδοι δέ ύπέκυψαν Πέρσησι διά την τούτου πικρότητα, push άρξαντες
της άνω "Αλυος ποταμού Άσίης έπ’ έτεα τριήκοντα καί εκατόν δυών
δέοντα, παρέξ ή όσον οί Σκύθαι ηρχον. push ύστέρω μέντοι χρόνφ
μετεμέλησέ τέ σφι ταΰτα ποιήσασι κ α ι άπέστησαν από Δαρείου·
άποστάντες δέ όπίσω κατεστράφησαν μάχη νικηθέντες. Ρ0Ρ τότε δέ έπί
Αστυάγεος οί Π έρσαι τε καί ό Κύρος έπαναστάντες τοΐσι Μήδοισι
ήρχον τό άπό τούτου της Άσίης. (1.130.1-2)
‘The Medes surrendered to the Persians because of Astyages’ harsh rule;
push they had been masters of Asia beyond the Halys for a hundred and
twenty-eight years except forthe period of Scythian domination. PUsh At a
later period they regretted their submission and revolted from Darius, but
were defeated and again reduced. P0P On the present occasion, under
Astyages, the Persians under Cyrus rose against the Medes and from then
onwards were masters of Asia.’

It might reasonably be expected that once having surrendered the Medes would stay
loyal subjects of thè Persian kings, but this expectation is denied in the embedded
sequence, which refers to events which it so happens we can date rather precisely to
thirty years after the Persian take-over. There is Discourse Topic continuity between
the sentences before and after the PUSH; after the POP, by contrast, the Medes have
vanished from the narrative. There is, logically, also some thematic continuity in the
embedding sequence: apart from the anaphoric τότε, which brings us back to the time
of the main story-line, there is also the repeated ‘masters of Asia’.
But it is also possible to interpret this passage as d e d i c t o : one might have
expected that Herodotus would continue the main story-line; instead, he inserts fac­
tual information about things that happened later.44

44 Sim ilarly 5.68.2 an d perhaps 9.37.4.

 S-R. SLIN G S

3.4. PUSH particle, d e d ic to

Anyway, in most instances, the denial of expectation is operative at the d e d ic to level,

as it is for μέντοι as a POP particle in (24), (25) and (26):

(32) αύτίκα δέ έπεμπε θεοπρόπους ές των έξηγητέων Τελμησσέων.

άπικομένοισι 5έ τοΐσι θεοπρόποισι και μαθοΰσι προς Τελμησσέων τό
θέλει σημαίνειν τό τέρας, ούκ έξεγένετο Κροίσω απαγγεΐλαν push
πριν γάρ ή όπίσω σφέας άναπλώσαι ές τάς Σάρδις ήλω ό Κροΐσος. push
Τελμησσέες μέντοι τάδε έγνωσαν (explanation o f the miracle). POP
Τελμησσέες μέν νυν ταΰτα ύπεκρίναντο Κροίσω ήδη ήλωκότι, οΰδέν
κω είδότες των ήν περί Σάρδις τε καί αυτόν Κροΐσον. (1.78.2-3)
‘At once he sent messengers to the interpreters of Telmessus. His messen­
gers did indeed arrive at Telmessus and learn the significance of the omen
from them, but they never managed to tell Croesus what it was— push f°r
before their ship made Sardis on the return voyage, Croesus had been taken
prisoner, push The interpretation of the men from Telmessus was th at... P0P
The men of Telmessus gave this answer to Croesus, who had already been
made a prisoner, though they didn’t have news from Sardis or Croesus yet.’

The story might seem over with Croesus being taken prisoner. But Herodotus still
wishes to record the explanation of the omen, even though it is irrelevant by now.
This is a footnote, and the footnote is meant for the reader. As in (26), there is no
adversativity if the text is read in a linear way, but PUSH/POP analysis shows that
one reads texts in that way at one’s peril.
A number of times, the contrast is between two parallel events, one taking place at
the main narrative moment, the other later:

(33) (After the battle of Plataea, Pausanias finds huge quantities of gold and silver,
left behind by the Persians. He comments on this to the Greek commanders:)
ταΰτα μέν Παυσανίην λέγεται είπ εΐν προς τούς στρατηγούς των
'Ελλήνων, push ύστέρφ μέντοι χρόνφ μετά ταΰτα καί των Πλαταιέων
ευρον συχνοί θήκας χρυσού καί αργύρου κ α ί των άλλων χρημάτων.
(9.82.3-83.1)
‘This is what Pausanias is reported to have said to the Greek commanding
officers, push After these events many people in Plataea, too, found coffers
full of gold, silver, and other valuables.’45

The reader’s expectation that the main story-line will continue is contradicted. This
analysis holds good as well for the next example, where ‘modification’ better

45 T h e em bedded stre tch g oes on w ith o th e r discoveries la ter m a d e b y the Plataean s. T h e P O P

occurs at 84.1.
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 121

accounts for how μέντοι PUSHes work than ‘denial of expectation’. They have to
do with variant versions of the same story:

(34) έπ είτε δε καί ταΰτα έκαμον ποιεΰντες (push ° YÒp δή νεκρός άτε
τεταριχευμένος άντεΐχέ τε καί ούδέν διεχέετο), Ρ0Ρ έκέλευσέ μιν ό
Καμβύσης κατακαΰσαι, έντελλόμενος οΰκ όσια push (explanation why
this is sacrilegious both for Persians and for Egyptians) pop out© δή ούδε-
τέροισι νομιζόμενα ένετέλλετο π οιέειν ό Καμβύσης. push μέντοι
Αιγύπτιοι λέγουσι, ούκ ’Άμασις ήν ό ταύτα παθών, άλλά άλλος τις τών
Αιγυπτίων έχων την αυτήν ήλικίην Α μ ά σ ι... (3.16.2-5)
‘When at last the executioners were weary, push f°r Amasis’ body had been
embalmed and would not fall to pieces under the blows, POp Cambyses
ordered it to be burnt. This was a wicked thing to order PUsh - p o p Camby­
ses, therefore, in giving this order, was running counter to the ritual practices
of both nations. p u ,s h The Egyptians have a story that it was not Amasis at all
whose body received this treatment; it was another man’s, of about the same
stature...’

The story as reported by Herodotus is not contradicted by the Egyptians: it is modi­

fied in one detail. I don’t see that it is useful to look for a d e r e expectation that is
denied by μέντοι, unless it is Cambyses’ expectation that it is Amasis’ corpse which
he is maltreating, but there is no previous signal in the text that there may be doubt as
to the identity of the body—in fact after the POP Herodotus goes on to record his
disbelief in this Egyptian version.46

3.5. C onclusions

Like άλλά, μέντοι shows its prototypical use (as a denial-of-expectation adversative)
more in PUSHes and POPs than in parallel sequences, where it has a considerable
overlap with δέ, in that it is also found in complex contrasts (often after a preceding
μέν).
At a first glance, μέντοι seems to be a particle that works much the same way as
Latin a t , to which Kroon (1995: 362) attributes ‘frustration of expectation’ as basic
meaning (cf. Sicking 1993: 34). However, there is a fundamental difference. A t is
typically an interactional particle—in Herodotus, μέντοι is no more interactional than
άλλά.47 It is true that μέντοι is relatively often used d e d i c t o compared to άλλά and
καίτοι but when it is, it is more often used, in Kroon’s terminology, at the presenta­
tional than at the interactional level—the explanation for the frequency of the d e d ic to

46 O ther instanc es o f μ έ ν τ ο ι used

4.109.1; 5.87.2; 6.66.3; 8.85.1.
dedicio in P U S H es are: 1.13.2; 2 .83; 3 .9 6 .1 ; 137.5; 149;

47 T his sta te m ent should b e read w ith the proviso m ade in n o te 13.
 S.R. SLINGS

use must lie elsewhere.48 μέντοι could never be the Greek equivalent of at in
Kroon’s very first example (1995: 333 (1)):

(35) (PI. C a s . 911)

E l o q u e r e .— a t p u d e t
Go on, tell me.—But I’m ashamed to.

I feel pretty certain that αλλά is the only particle that can be used here (Denniston
1954: 7).

4. καίτοι

Unlike άλλά and μέντοι, καίτοι is never used within the clause or sentence.49 This
deprives us of one possible clue as to its function within larger discourse units. Yet
the picture is clear enough. (30) schematises how μέντοι works atPUSHes; καίτοι
is its mirror image. In many studies on adversative (and causal) relators it is assumed
that there is a basic order between the two conjuncts related by them (‘although A,
B ’; ‘A but B ’; cf. Sanders e t a l. 1992; Torek 1996). As a PUSH particle, καίτοι is
used only for non-basic order:

(36) A PUSH καίτοι B (A denial exp. B)

In other words, after an A statement has been made, a B statement is made which
would lead one to believe that the A statement, although true (this is what μέντοι and
καίτοι have in common as against άλλά), is incompatible with what one would have
inferred given the truth of B. (καίτοι is, I think, always about truths of statements,
not about desirability or otherwise of courses of action.) Put briefly, the difference is
between ‘He is rich μέντοι he is unhappy’ and ‘He is unhappy καίτοι he is rich’.
As a matter of fact, (36) is pretty close to a postponed concessive clause ‘A, although
B ’— the difference being that καίτοι, like μέντοι, puts the emphasis o n th e B con­
junct. I will call the relation symbolised by (36) ‘inverted denial of expectation’.
Greek may well be special inasmuch as it has a separate particle for this relation.
Such a relator has not been identified for Latin or any of the major West-European
languages. An English equivalent would be the cumbersome phrase a n d th a t d e s p i t e
th e f a c t th a t. Yet a word of warning is in order: the value as symbolised in (36) is
based only on a PUSH/POP analysis— in parallel sequences, καίτοι seems to
behave differently altogether.

48 If I may venture a guess, the scarcity of d e dicio use of δέ may have played a part. But since I
have not studied δέ from the PUSH/POP angle, I have to leave it at that.
49 The value of καίτοι proposed here, 'inverted denial of expectation’, would make καίπερ its
equivalent within the sentence, of course only when it qualifies participles that follow the main
verb. It is interesting to note that from the fourth century onwards καίτοι tends to replace καίπερ
(Denniston 1954: 559), and seems to have replaced it altogether in late Hellenistic Greek
(Blomqvist 1969: 41-43), At the de dicto level, Herodotus uses καί ταΰτα instead of καίπερ
(2.120.4;6.61.2).
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 123

4.1. d e re

The inverted denial of expectation is best illustrated by a group of passages in which

the A statement refers to failure to achieve something and the καίτοι PUSH to the
effort made to achieve it. An effort is not a denial of an expectation raised by a fail­
ure—on the contrary, failure presupposes effort, Rather, failure is a denial of an
expectation raised by an effort:

(37) των δε βαρβάρων ό πεζός υπό την παρεοΰσαν νύκτα έπορεύετο έπ ί την
Πελοπόννησον, posh καίτοι τά δυνατά πάντα έμεμηχάνητο οκως κατ'
ήπειρον μή έσβάλοιεν οί βάρβαροι. (8.71.1)
‘And the Persian army that very night was on the march for the Peloponnese.
PUSH Nevertheless everything that ingenuity could contrive had been done to
prevent the Persian army from forcing the Isthmus.’

One would have expected that the Persians were not going to march against the Pelo­
ponnesus (the contrary of the A statement), given' the efforts described in the embed­
ded sequence.50 Note the Discourse Topic continuity (cf. the remark made above on
μεντοι PUSHes).
It is in the nature of καίτοι as a PUSH particle, with its non-basic order, that it
tends to trigger a POP which restates the A conjunct. When the PUSH raises an
expectation which had been denied previously, coherence may dictate that the denial
of expectation should be repeated:

(38) επτά δέ μηνών καί ένιαυτοΰ διεληλυθότος ήδη ό Δαρειός τε ήσχαλλε

καί ή στρατιή πάσα ού δυνατή έοΰσα έλ εΐν τούς Βαβυλωνίους, push
κα ίτοι πάντα σοφίσματα καί πάσας μηχανάς έπεποιήκεε ές αυτούς
ΔαρεΧος' pop άλλ' οΰδ' ώς έδύνατο έλεΧν σφεας ... (3.152)
‘A year and seven months went by, and Darius and his army began to chafe
at their inability to make any progress towards taking the city. PUsh Every
trick, every possible device, had been tried; pop but to no purpose. The town
could not be taken.’

The Discourse Topic is repeated literally, and it retains its status after the POP,
because of the repetition of the A conjunct. For the use of άλλά at the POP, cf. the
remarks made about (16) and (20). An example outside the failure - effort pattem:

(39) τούς δέ προς τά λεγάμενα ύποκρίνασθαι ώς έτοιμοί ε ίσ ι Ά ργεΐοι

π οιέειν ταΰτα τριήκοντα έτεα ειρήνην σπεισάμενοι Λακεδαιμονίοισι
καί ήγεόμενοι κατά το ήμισυ πόσης της συμμαχίης· push καί-τ°ι κατά

50 The POP is not found until 74.1. Instances outside the pattern of failure - effort: 2.148.2; 3.15.3;
8.86 (related to failure - effort).
124 S.R. SLINGS

γε to δίκαιον γίνεσθαι την ήγεμονίην έωυτών, ΡΟρ άλλ' όμως σφι

άποχραν κατά τό ήμισυ ήγεομένοισι. (7.148.4)
‘The Argive answer was, that they were willing to do what they were asked
upon two conditions: first they must obtain a thirty years’ truce with Sparta,
and, secondly, share with Sparta, on equal terms, the c o m m a n d of the confed­
erate forces, push By right Argos was entitled to the sole command; pop
nevertheless they would be content with an equal division.’51

Note the thematic continuity throughout: ήγεόμενοι... ήγεμονίην ... ήγεομένοισι.

In some cases, the opposition between the A and B statements is more direct in
that B expresses the expectation itself:

(40) δύο δ' εχων ταΰτα έχει πάσαν κακότητα· PUSH τα μέν γάρ ϋβρι κεκο-
ρημένος έρδει πολλά καί ατάσθαλα, τά δέ φθόνω. push Κ{*ίτοι άνδρα
γε τύραννον αφθονον έδει είναι, έχοντά γε πάντα τά αγαθά' Ρ0Ρ τό δέ
ύπεναντίον τούτου ές τούς πολιήτας πέφυκε- (3.80.4)
“‘These two vices are the root cause of all wickedness in the monarch: push
by pride and by envy he is led to acts of savage and unnatural violence. Pu$h
Absolute power ought, by right, to preclude envy on the principle that the man
who possesses it has also at his command everything he could wish for; POP
but in fact it is not so, as the behaviour of kings to their subjects proves.”’

Here A is the denial of an expectation not raised by B but expressed in B. The coun-
terfactuality of the καίτοι sentence,52 and the fact that it deals only with envy, not
with pride, show that the καίτοι clause is an embedded sequence. Once again, the
POP is used to repeat the A conjunct.

4.2. d e d ic to

At the d e d ic to level, καίτοι is used when a view is stated, and a contrary view, which
is not accepted, treated in an embedded sequence:

(41) και άριστος έγένετο μακρω 'Αριστόδημός κατά γνώμας τάς ήμετέρας ...
μετά δέ τούτον ήρίστευσαν Ποσειδώνιός τε ... push κοίτοι γενομένης
λέσχης ος γένοιτο αύτών άριστος, έγνωσαν οί παραγενόμενοι
Σπαρτιητέων 'Αριστόδημον μεν βουλόμενον φανερώς άποθανεΐν έκ της
παρεούσης οί αίτίης, λυσσώντά τε καί έκλείποντα τήν τάξιν έργα
άποδέξασθαι μεγάλα, Ποσειδώνιον δέ ού βουλόμενον άποθνήσκειν
άνδρα γενέσθαι αγαθόν τοσούτω τούτον είν α ι άμείνω. Ρ0Ρ άλλα
ταύτα μέν καί φθόνω αν ε ίπ ο ιεν (9.71.2-4)

51 Cf. 7.9β.1 (POP ai 9β.2 τοίνυν).

52 Cf. 5.45.2. With different signals: 7.10γ.2.
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 125

‘Much the greatest courage was shown, in my opinion, by Aristodemus,...

After him, the greatest personal distinction was won by three Spartans, Posi­
donius ... puSHHowever, when, after the battle, the question of who had most
distinguished himself was discussed, the Spartans present decided that
Aristodemus had, indeed, fought magnificently but that he had done so ... in
his desire to be killed in his comrades’ eyes; Posidonius, on the contrary,
without any wish to be killed, had fought bravely, pop It may, of course, have
been envy which made them say this;'

Herodotus states his personal view and maintains it in spite of the fact that the Spar­
tans present at Plataea thought differently. One might have expected him to go by the
decision of those who had been there, but a denial of this expectation is expressed in
the A conjunct. The POP dismisses the Spartans’ dissenting view definitively.53
There is one instance of this use in a speech:

(42) Ώ βασιλεύ, άρχήθεν ήπιστάμην οτι άληθείη χρεωμένος ού φίλα τοι

έρέω. συ δέ έπεί ήνάγκασας λέγειν των λόγων τούς αληθέστατους,
έλεγον τά χατήχτοντα Σπαρτιήτησι. Ρ καίτοι ώς έγώ τυγχάνω τα νυν
τάδε έστοργώς εκείνους, αυτός μάλιστα έξεπίστεαι, pÜSH “Π-Ρήν
τε καί γέρεα άπελόμενοι πατρώια άπολίν τε καί φυγάδα πεποιήκασι,
πατήρ δέ <ό> σός ύποδεξάμενος βίον τέ μοι καί οίκον έδωκε. Ρ0Ρ οΰκ
ών οίκός έστι άνδρα τόν σώφρονα εύνοίην φαινομένην διωθέεσθαι,
αλλά στέργειν μάλιστα. (7.104.1-2)
‘“My lord, I knew before I began that if I spoke the truth you would not like
it. But, as you demanded the plain truth and nothing less, I told you how
things are with the Spartans, push Yet you are well aware that I now feel but
little affection for my countrymen, PUsh wh° robbed me of my hereditary
power and privileges and made me a fugitive without a home—whereas your
father welcomed me at his court and gave me the means of livelihood and
somewhere to live, pop Surely it is unreasonable to reject kindness; any sen­
sible man will cherish it.’”

Despite the fact that D e m a r a tu s has been treated badly by the Spartans and kindly by
Darius, he gives his objective opinion about the Spartans, which Xerxes might not
have expected from him.

5. Conclusions

The main conclusion from this survey is that PUSH/POP analysis reveals that the
particles that I have studied behave quite differently in hierarchical sequences than in
non-hierarchical ones. Their PUSH/POP use is similar to their intra-sentential or
even intra-clausal use (except for καίτοι, which has no such use). Apparently,

53 Cf. 4.77.1 (POP at 77.2).

 S.R. SL IN G S

PUSH/POP contexts are on the whole more similar to intra-sentential contexts than
initial-sentence contexts in parallel sequences.
There is, on the other hand, an important difference between PUSH and POP
particles. PUSH particles mark a relation between the preceding central conjunct and
the following subsidiary one—as such they are not different from panicles used in
parallel sequences. By contrast, POP particles tend to ignore the preceding embedded
sequence—normally when used d e r e , less often when used d e d ic to —and to relate
two central conjuncts, sometimes at a considerable distance. Therefore, any analysis
that leaves out the hierarchy of the text is bound to be deficient.
Furthermore, these particles are much more clearly adversative than is often
assumed in recent studies, and each in their separate ways: αλλά replacing, μέντοι
denial of expectation, καίτοι inverted denial of expectation. This becomes even
clearer when the distinction between d e r e and d e d ic to use is taken into account for
adversative particles, as well as for causal particles, where it is more or less common­
place. I hope that I have shown that the distinction is in fact vital for the study of
adversatives.
Of course, I cannot possibly claim th a t a d v e r s a tiv ity —whatever th a t m a y m e a n —
should be written into the basic meaning of the three particles that I’ve analysed.
PUSH/POP analysis is only a small part of Discourse Analysis. On the other hand,
the results discourage a hasty writing-off of adversativity as a side effect of a deeper
basic value of these Greek particles. I am convinced that Kroon’s treatment of adver­
sative particles in Latin has established that adversativity is, for Latin particles, indeed
such a side effect,54 but given the present state of our knowledge of Greek particles, I
am not prepared yet to assume that Greek and Latin particles are similar in this
respect.
Indeed, when I observe that in the most recent publications on Greek particles,
two scholars whom I respect very much. Dr Wakker and Prof. Sicking, are in funda­
mental disagreement on important words like αρα and δή,55 and when I have to say,
reluctantly, that both of them seem to me fundamentally unsound on γε,561feel that
a moratorium on basic values for Greek particles would, in the current state of affairs,
be highly desirable. There is a lot of work to be done in the way of detailed analysis,
and there is a lot to be learned from the study of particles in the living West-
European languages. Only through the painstaking examination of various types of
contexts, and through the comparison of Greek particles with particles in other lan­
guages, can we really advance. Until then, making pronouncements on basic values
amounts to little more than armchair linguistics—it is hardly more helpful than the

54 Even this is not entirely true. ‘Frustration of expectation’. Kroon’s basic value of at, is one basic
form of adversativity. And Kroon does not deal with sed, which may well be adversative basically.
55 Sicking (1986: 133) and Wakker (1994: 343f.) on άρα (cf. Van Ophuijsen 1993: 82); Sicking
(1993: 52) and Wakker (1994: 351) on δή.
56 Sicking (1986: 125) and Wakker (1994: 308) describe ye as a kind of Focus particle, an
analysis that cannot do justice to its use in adding constituents to already complete sentences (quite
apart from the fact that γε hardly ever accompanies the true Focus of a sentence).
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 127

practice of most classical scholars, for whom the be-all and end-all of the study of
Greek particles is the ability to find the correct page in Denniston.57

Bibliography

Anscombre, J.C. & O. Ducrot

1977 ‘Deux m a i s en français?’. L i n g u a 43, 23-40
Bakker, E.J.
1993 ‘Boundaries, Topic and the Structure of Discourse: An Investigation of the
Ancient Greek Particle d é '. S tu d ie s in L a n g u a g e 17,275-311
Blomqvist, J.
1969 G r e e k P a r tic le s in H e lle n is tic P r o s e . Lund
Denniston, J.D.
19542 T h e G r e e k P a r tic le s . Oxford
Drescher, M. & Th. Kotschi
1988 ‘Das ‘Genfer Modell’. Diskussion eines Ansatzes zur Diskursanalyse’.
S p r a c h e u n d P r a g m a tik 8,1 -42
Foolen, A.P.
1993 D e b e te k e n is v a n p a r t i k e l s : E e n d o k u m e n t ä r e v a n d e s t a n d v a n h e t o n d e r ·
z o e k m e t b ijz o n d e r e a a n d a c h t v o o r maar. Nijmegen
Groningen, B.A. van
19592 H e r o d o t u s ’ h is to r ie n m e t in le id in g e n c o m m e n ta a r III. Leiden
Halliday, M.A.K.
1985 A n I n t r o d u c tio n to F u n c tio n a l G r a m m a r . London-Baltimore
Jefferson, G.
1972 ‘Side Sequences’. In: D. Sudnow (ed.). S t u d i e s in S o c i a l I n t e r a c t i o n . New
York, 294-338
Kroon, C.H.M.
1989 ‘Causal Connectors in Latin: The Discourse Function of n a m , e n im , ig i t u r ,
and e r g o . In: M. Lavency & D. Longrée (eds). A c t e s d u c in q u iè m e C o llo q u e
d e L in g u is tiq u e la tin e . Louvain-la-Neuve, 231-243
1995 D i s c o u r s e P a r t i c l e s i n L a t i n : A S tu d y o f nam, enim, autem, vero a n d at.
Amsterdam
1996 ‘Discourse Markers, Discourse Structure and Functional Grammar’. In:
C.S. Butler, J.H. Connolly, R.A. Gatward & R.M. Vismans (eds),
D i s c o u r s e a n d P r a g m a tic s i n F u n c tio n a l G r a m m a r . Berlin-New York, 17-
32

57 I thank my friends and colleagues Ad Foolen (Catholic University Nijmegen), Caroline Kroon
(Free University at Amsterdam), Gisela Redeker (ditto) and Rodie Risselada (University of
Amsterdam) for their helpful comments on, and fundamental criticism of, various preliminary ver­
sions of this paper.
 S.R. SLINGS

Lakoff, R.
1971 ‘If’s, And’s and But’s about Conjunction’. In: C J. Fillmore & D.T.
Langendoen (eds), Studies inLinguistic Semattiics. New York, 114-149
Ophuijsen, J. van
1993 ΌΥΝ, APA, ΔΗ, TOINYN: The Linguistic Articulation of Arguments in
Plato’s Phaedo’. In: C.M.J. Sicking & J.M. van Ophuijsen, Two Studies in
Attic Particle Usage. Leiden, 69-164
Polanyi, L.
1988 ‘A Formal Model of the Structure of Discourse’. Journal o f Pragmatics 12,
601-638
Polanyi, L. & R.J.H. Scha
1983 ‘The Syntax of Discourse’. Text 3, 261-270
Powell, J.E.
19382 A Lexicon to Herodotus. Cambridge
Roulet, E.
1991 ‘Vers une approche modulaire de l ’analyse du discours’. Cahiers de lin-
guistique frangaise 11, 53-81
Roulet, E. et al.
1985 L’articulation du discours en frangais conlemporain. Bern
Ruijgh, C.J.
1971 Autourde 'τε épique’ : Études sur la syntaxe grecque. Amsterdam
Sanders, T.J.M., W.P.M. Spooren &L.G.M. Noordman
1992 Towards a Taxonomy of Coherence Relations’. Discourse Processes 15,1-
35
Schiffrin, D.
1985 ‘Multiple Constraints on Discourse Options: A Quantitative Analysis of
Causal Sequences’. Discourse Processes 8, 281-303
Sélincourt, A. de
1972 Herodotus: the Histories, revised, with an Introduction and Notes by A.R.
Bum. Harmondsworth
Sicking, C.M.J.
1986 ‘Griekse partikels: definitie en classificatie’. Lampas 19,125-141
1993 ‘Devices for Text Articulation in Lysias I and XII’. In: C.M.J. Sicking &
J.M. van Ophuijsen, Two Sudies in Attic Particle Usage. Leiden, 1-66
Slings, S.R.
1980 ‘KAI adversativum: Some Thoughts on the Semantics of Coordination’. In:
D.J. van Alkemade et al. (eds), Linguistic Studies Offered to Berthe
Siertsema. Amsterdam, 101-125
Spooren, W.P.M.S.
1989 Some Aspects o f the Form and Interpretation o f Global Contrastive
Coherence Relations. Nijmegen (diss. KUN)
de Strycker, E. & S.R. Slings
 ADVERSATIVE RELATORS BETWEEN PUSH AND POP 129

1994 P l a t o ’s A p o l o g y o f S o c r a t e s . A L i t e r a r y a n d P h i l o s o p h i c a l S t u d y w i t h a
R u n n i n g C o m m e n ta r y . Leiden
Sweetser, E.
1990 F rom E t y m o l o g y to P r a g m a t i c s . M e t a p h o r i c a l a n d C u l t u r a l A s p e c t s o f
S e m a n t i c S t r u c tu r e . Cambridge
Torek, D.
1996 A s p e c t s d e la c a u s a l i t à d i s c u r s i v e e n f r a n g a i s o r a l c o n te m p o r a in . Amster­
dam (diss. Free University)
Wakker, G.C.
1994 C o n d i t i o n s a n d C o n d itio n a ls : A n I n v e s t i g a t i o n o f A n c ie n t G re e k . Amsterdam
 SUR LE RÔLE PRAGMATIQUE DE K A IT O I

B e r n a r d ja c q u in o d
Université de Saint-Étienne

1. Introduction

La particule καίτοι n’est pas attestée dans nos plus anciens textes grecs. Aussi le
professeur C J . Ruijgh n’en parle-t-il dans Autour de τε épique que pour dire qu’elle
ne se trouve pas dans la vulgate homérique, à l’exception d’une variante qu’il rejette
(1971: § 735). Ce statut m’est apparu comme un avantage, et l’étude de καίτοι en
devrait être simplifiée par l’absence d ’une très longue histoire avant les textes clas­
siques. Néanmoins, J.D. Denniston a subdivisé la présentation des emplois de ce mot
en 14 paragraphes, ce qui paraît beaucoup. Le plus troublant est que l’on attribue des
sens opposés à ce mot récent. Ainsi J. Humbert (1954: 415) déclare-t-il qu’elle «est à
la fois intensive et restrictive». La syntaxe de Kühner-Gerth (2, 151-2) affirme que
καί-rot sert à opposer (entgegenstellt), puis avoue que la particule apparaît aussi sans
qu’il y ait opposition (ohne Gegensatz). Mais c ’est le sens d’opposition qui est le
plus souvent retenu comme le plus important.1
J.D. Denniston, en Introduction à son livre, admet que καίτοι, comme une série
d’autres particules, est utilisée «both to add and to contrast» (1954: xlix). Et il pré­
cise: «custom attaches an adversative force to a pure connective like καίτοι» (1954:
xlix). Mais il ajoute en note: «Adversative καίτοι is so common that one may legiti­
mately reckon an adversative sense as one of the senses of the particle. On the other
hand, where κ α ί appears to be adversative, the opposition is inherent in the context
rather than expressed by the particle. Decision between these two explanations is
often a delicate matter. We should not, I think, resort to the second in the case of
well-established usages. Thus I cannot agree with des Places when he says (p. 107)
that in corrective μέν οΰν ‘l'opposition réside uniquement dans la pensée, non dans
la particule’».
Finalement, d’un point de vue synchronique, Denniston situe la particule καίτοι
dans une position intermédiaire entre les eliminative adversatives et les balancing
adversatives,2 ce qui est intéressant.

1 Par exemple, la syntaxe de O. Riemann - Ch. Cucuel, Règles fondamentales de la syntaxe

grecque4, Paris, 1901, p. 238, la range dans les particules adversatives.
2 «Intermediate between the two [= between eliminative adversatives and balancing adversatives]
we have adversatives tike άτάρ and καίτοι (and sometimes αλλά) which raise an objection, leaving
it uncertain whether the objection is a fatal one or not» (1954: xlix).
 BERNARD JACQUINOD

D ’un point de vue diachronique, Denniston, dans son introduction au chapitre con-
sacré à καίτοι, suppose un sens connectif à l’origine, à cause du sens des Constitu­
ante (= «and, I would have you know», 1954: 555-6). Mais il écrit ensuite «The con­
nective sense, though prima facie the earlier, makes its appareance, in fact, later than
the adversative : while with καί μην the contrary is the case». Il précise en outre
«There is usually a certain combative tone in καίτοι. For this reason it is not com­
mon in unimpassioned, cold-blooded exposition. It is significant that out of 24
Thucydidean examples all except i 10.2 are from speeches (viii 72.1 reported
speech)».
Son plan est fondé sur trois valeurs principales (adversative, continuative, logi­
cal), avec une quatrième partie consacrée aux alliances de particules, avec chaque fois
plusieurs paragraphes pour rendre compte de la variété des emplois:

(1) Adversative
(1) In general
(ii) Used by a speaker in pulling himself up abruptly
(iii) The objection introduced by καίτοι is countered, palliated, or modified by a
following adversative clause
(iv) A variant of the above is the forecasting of the following adversative by μέν
(v) ... used at the opening of a speech
(vi) καίτοι in parenthesis
(vii) ... a tendency ... to develop from an adversative particle introducing a main
clause into a concessive particle introducing a subordinate clause

(2) Continuative

(3) Logical, καίτοι often marks the transition from premise to premise, almost
invariably from minor to major, only very rarely vice versa
(i) Occasionally in a complete syllogism, with conclusion expressed
(ii) Far more frequently the conclusion of the syllogism is left to the imagination
(iii) If the relation of the καίτοι sentence to the preceding sentence is regarded in
isolation, apart from the role played by both in a syllogistic structure, it is
usually adversative in tone.

(4) καίτοι combined with other particles

(i) κ α ίτοιγε, κ α ίτ ο ι... γε
(ii) καίτοι περ
(iii) In A. E u . 849 neither καίτοι μέν (Μ) nor καίτοι γε μην (F I , Fa) is possible.

Je pourrais faire de nombreuses remarques sur ce plan, mais mon but n’est pas de
critiquer qui que ce soit, et je sais combien mes devanciers ont dit des choses justes
sur ce mot. Je voudrais seulement chercher s’il peut etre possible de trouver une unité
aux emplois de cette particule (comme Ta fait le Professeur Ruijgh pour d’autres
 SUR LE RÔLE PRAGMATIQUE DE ΚΑ1ΤΌ Ί 133

particules), ce que semble autoriser la récente création de καίτοι, et mieux formuler,

peut-être, ce que d’autres ont dit avant moi, mieux répartir ce qui revient à la particule
et ce qui se déduit du contexte, peut-être déplacer le point du vue: au lieu de chercher
à situer d’abord la particule dans une taxinomie (est-elle ou non adversative?),
chercher d’abord quelle est pragmatiquement sa fonction.

2. L ’em ploi de κ α ίτ ο ι chez P la to n

2.1. L a v ale u r de κ α ίτ ο ι

H m’est apparu que Platon offrait un corpus privilégié par l ’ampleur de l’emploi de
cette particule, qui se rencontre 140 fois chez cet auteur, et par la variété au moins
apparente de son usage, puisque Denniston cite Platon dans 12 des 14 rubriques
consacrées à καίτοι. Platon a l ’avantage de présenter aussi bien des récits que des
dialogues, des échanges faits de répliques courtes, pleins de vivacité, à côté de pas­
sages construits comme de véritables petits exposés. C’est sur cet auteur que, dans
un premier temps, j ’ai concentré mes efforts, non sans regarder ensuite un peu
ailleurs. A l ’usage il me semble que la présence d’un γ ε ne modifie pas la valeur de
καίτοι, et j ’ai pas cru bon de distinguer les exemples avec καίτοι seul de ceux avec
κ α ίτο ι... γε.

En ce qui concerne καίτοι, le corpus platonicien se révèle finalement d ’une assez

grande homogénéité. Cette particule est employée chez Platon dans des raison­
nements. Elle introduit un élément nouveau, non prévisible de façon évidente d’après
le contexte (une incidente), qui doit relancer la réflexion.
Elément nouveau qui doit relancer la réflexion, cela peut se trouver sous la forme
d’une opposition:

Τφ δεδιέναι αρα καί δ έει άνδρειοί είσι πάντες πλήν οί φιλόσοφον

κ α ί τ ο ι αλογόν γ ε δ έει τινά καί δειλίςι άνδρείον είναι. (Plat. P h d .
68dll-12)
«ainsi, sauf chez les philosophes, c’est la crainte et la peur qui rend tous les
hommes courageux. Et pourtant, n’est-il pas illogique d ’être courageux par
peur et lâcheté?» (trad. L. Robin)

La contradiction peut être apportée à soi-même:

μετά δ’ 'Εστίαν δίκαιον "Ρέαν καί Κρόνον έπισκέψασθαι. κ α ί τ ο ι τό γ ε

του Κρόνου όνομα ήδη διήλθομεν. ’Ίσως μέντοι ούδέν λέγω. (Plat. C r a t .
401el-3)
«Après Hestia, il est juste d’examiner Rhéa et Kronos. Le nom de Kronos, à
vrai dire, nous l’avons déjà passé en revue. Mais peut-être mes propos sont-
ils sans valeur.» (trad. L. Méridier)
 BERNARD JACQUINOD

Parfois on aplus l’impression d’une pensée qui continue son cheminement:

ΣΩ. Τί οΰν δή τούτο λέγεις;

ΓΟΡ, Ίο π είθ ειν εγωγ’ οΐόν είν α ι τοίς λόγοις κ α ί έν δικαστηρίφ
δικαστάς καί έν βουλευτηρίφ βουλευτάς καί έν έκκλησί? έκκλησι-
αστάς καί έν άλλω συλλόγω παντί, οστις άν· πολιτικός σύλλογος
γίγνηται. κ α ίτ ο ι έν ταύτη τη δυνάμει δοΰλον μεν εξεις τόν ιατρόν,
δοΰλον δέ τον παιδοτρίβην (Plat. G r g . 452d8-e5)
«Socrate: Mais enfin, quelle est donc cette chose dont tu veux parler?
Gorgias: Ce dont je veux parler, c’est du pouvoir de persuader par le discours
les juges au tribunal, les sénateurs au Conseil, le peuple dans l ’Assemblée du
peuple et de même dans toute autre réunion qui soit une réunion de citoyens.
Or, avec ce pouvoir, tu feras ton esclave du médecin, ton esclave du
pédotribe».

Denniston décrit cet exemple comme un syllogisme sans conclusion. Comme exem­
ple purement c o n t i n m t i f , il cite

Πάνυ μέν οΰν, εφη, εοικε τούτω ώ λέγεις· κ α ίτο ι γ' έν τη ήμετέρα
πόλει τους έπικούρους ώσπερ κύνας έθέμεθα υπηκόους των αρχόντων
ώσπερ ποιμένων πόλεως. (Plat. R e s p . 440d4-6)
«Hé ! oui, fit-il, avec ce que tu dis la ressemblance est parfaite. Dans la consti­
tution de notre Etat, nous avons d ’ailleurs fait de nos auxiliaires des manières
de chiens, dociles à la voix de chefs, qui sont comme les bergers d’une cité»
(trad. Robin)

et

άθρει ότι οΰδέ παναληθής έστιν ή των άλλων ηδονή πλήν της του φρονί­
μου ουδέ καθαρά, άλλ’ έσκιαγραφημένη τις, ώς έγώ δοκώ μοι των
σοφών τίνος άκηκοέναι. κ α ίτο ι τοΰτ αν ε'ίη μέγιστόν τε καί κυριώτα-
τον των πτωμάτων. (Plat. R e s p . 583b3-7)
«Considère qu’à part le plaisir du sage, le plaisir des autres n’est ni bien réel
ni pur, ce n’est qu’une ombre de plaisir, comme je crois bien l ’avoir entendu
dire à un sage; et s’il en est ainsi, ce pourrait bien être pour l’injuste la grande
chute, la chute décisive» (trad. E. Chambry).

De fait, ici encore, on ne voit pas de valeur proprement adversative. Toutefois, l’idée
introduite par καίτοι, tout en s’appuyant sur ce qui précède, amène une idée bien dif­
férente: d ’un plaisir imparfait, on passe à l ’idée de chute définitive. L’adjonction
modifie profondément ce qui précède.
 SUR LE RÔLE P R A G M ^ Q U E DE KAJTOI 135

Je pense qu’il faut revoir la présentation de Kühner-Gerth et ne pas traiter ce

type d ’emploi comme surprenant, marginal. H ne l’est que parce que la description de
la valeur de καίτοι n’a pas été posée correctement.

2.2. Le rôle pragm atique de κα ίτο ι

Mais cette première approche de καίτοι ne suffit pas, car elle vaudrait sans doute
pour de nombreuses particules ou associations de particules. Il convient d ’insister
sur son rôle dans l ’argumentation.
Certes, on ne peut oublier le rôle de notre particule dans les syllogismes, rôle qui
a été maintes fois souligné, comme on le voit en

Και μην, (δ Σώκρατες, ö γ ε έρωτφς περί επιστήμης ούκ αν δυναίμην

ώποκρίνασθαι ώσπερ περί του μήκους τε καί της δυνάμεως. κ α ίτ ο ι συ
γέ μοι δοκεις τοιοΰτόν τι ζη τειν ώ σ τε πάλιν αΰ φαίνεται ψευδής ό
Θεόδωρος. (Plat. T h t. 148b5-8)
«Et pourtant (Καί μην), Socrate, la question que tu me poses au sujet de la
science, je ne saurais la résoudre comme j ’ai fait celle qui a trait à la longueur
et la puissance. Or (καίτοι) c ’est bien, à ce que je crois, quelque chose
comme cela que tu cherches; voilà donc (ώστε), de nouveau, Théodore con­
vaincu de fausseté» (trad. A. Diès).

où καίτοι sert à introduire la mineure.

Denniston a de très bons paragraphes à ce sujet. Π me semble qu’il faut relever deux
choses. La première, c’est qu’il se demande si l’emploi dans les syllogismes provient
de la valeur adversative ou de la valeur connective de la particule. C’est le type même
du faux problème; d’abord, il faut partir de la valeur générale de la particule, ce qui
éviterait d ’avoir à conclure que cet emploi en logique provient des deux valeurs
(1954: 564), et, ensuite, il n ’y a aucune raison de faire un sort particulier à ce
domaine d ’emploi. Denniston note bien que καίτοι n ’est pas la particule la plus
traditionnelle dans les syllogismes et que καίτοι intervient aussi bien pour passer de
la majeure à la mineure que le contraire. A cela s’ajoute le fait que la conclusion est
souvent absente. De fait, dans les exercices d ’école, la seconde prémisse est intro­
duite par αλλά μην, par δέ γε, voire par δέ tout seul. On trouve de nombreux exem­
ples dans les F r a g m e n t s d e s S t o ï c i e n s . 3 Citons plutôt un exemple amusant dans
Lucien, J u p . t r a g . 51 εί γάρ είσί βωμοί, είσί καί θεοί' άλλα μην είσί βωμοί,
είσίν άρα κ αί θεοί.

Il faut donc décrire καίτοι comme un élément de l'argumentation en général, sans

s’obnubiler sur la notion de syllogisme. La présence de καίτοι dans un syllogisme
(canoniquement constitué, ou, ce qui est le cas le plus fréquent, restitué par les com-

3 Ioannes Ab Amim, Stoicorum veterumfragmenta, Stuttgart, 1964, vol. IV, Dialectica, p. 38-94.
 BERNARD JACQUINOD

mentateurs) n’est qu’un cas de son emploi dans un raisonnement, ce qui semble être
la fonction à quoi a été consacré καίτοι. Elle est notable, mais n’a pas à être analysée
à part, elle peut l ’être pour des raisons didactiques, non pour des raisons linguis­
tiques. On peut regretter le plan de Denniston, qui sépare adversative et logical

Quelle forme revêt la valeur adversative que nous décelons dans de très nombreux
cas? Pour mieux la cerner, partons d’un exemple. J’ai choisi la première phrase de
U Apologie de Socrate, mais cent autres auraient pu convenir:

"Ou μέν υμείς, ώ ανδρες 'Αθηναίοι, πεπόνθατε ύπό των έμών κατηγό­
ρων, οϋκ ο ΐδ α 1 έγώ δ’ οΰν κ α ί αυτός ύπ' αύτών ολίγου έμαυτοΰ
έπελαθόμην, οΰτω πιθανώς ελεγον. κ α ίτο ι αληθές γ ε ώς έπος είπέίν
ούδέν είρήκασιν. (Plat. Αρ. 17al-4)
«Je ne sais trop, Athéniens, quel effet mes accusateurs ont pu produite sur
vous. Pour moi, en les écoutant, j ’ai failli oublier qui je suis, tant leurs dis­
cours étaient persuasifs. Et pourtant, sans exagérer, ils n ’ont pas dit un seul
mot de vrai.»

Devant un tel exemple, il ne faut pas se demander si la particule est adversative ou

non (on peut soutenir l ’un et l ’autre), mais bien décrire son fonctionnement. Il peut
être éclairant de voir la portée argumentative de l’adverbe πιθανώς: cet adverbe situe
les propos des accusateurs de Socrate dans le domaine de la conviction, mais, d’un
point de vue argumentatif, la valeur persuasive peut être utilisée pour établir une
valeur de véridieité. De «il est convaincant», on est porté à conclure à «il dit vrai». La
phrase introduite par καίτοι ne récuse pas les propos précédents, n’annule pas l’idée
que les propos des accusateurs de Socrate sont tellement bien présentés qu’ils sont
capables de persuader, elle n’est donc pas véritablement adversative, mais elle élimine
une interprétation qui pourrait venir à l’esprit. Elle nous rappelle que celui qui per­
suade peut être menteur. Elle évite à l’interlocuteur de se fourvoyer sur une fausse
piste, de conclure, à partir de «il persuade», à «il dit vrai». On peut formaliser ainsi:

Nous avons une affirmation A (= il est convaincant) qui pourrait aller dans le
sens d’une conclusion c, non explicitée (= donc il dit vrai), καίτοι introduit une
affirmations (= en fait il a menti) qui ne détruit pas A (il reste qu’il est capable
de convaincre) (B Φ -A), mais qui pousse à conclure -c (= il dit des choses qui ne
sont pas vraies).

O. Ducrot a décrit d ’une façon proche un des deux emplois de la conjonction

française mais, celui qui correspond à all. aber et à espagnol pero (1975: 226-8).
Comme ce mais (et à la différence de all. sondern et esp. sino), καίτοι ne présup­
pose pas que la proposition précédente soit négative, ni de forme, ni de sens. Et on
peut ajouter que A et S n ’appartiennent pas toujours directement à la même classe
argumentative.
 SUR LE RÔLE PRAGMATIQUE DE KAITOI 137

2.3, E tu d e des différents em plois

Le classement des particules a peut-être été trop fait à partir d’une taxinomie qui se
voulait logique. H faudrait plus systématiquement rechercher leur rôle dans l ’argu­
mentation. H faut insister sur l ’emploi fortement pragmatique de ce marqueur qu’est
καίτοι. C’est un signal destiné à orienter la pensée de l’interlocuteur dans une direc­
tion différente. On peut reprendre des exemples représentatifs des différentes valeurs
accordées à καίτοι et voir si on peut les ramener à cette idée d ’un élément nouveau
dans le passage, nouveau au sens de non immédiatement déductible, qui incite à
infléchir le cours de la pensée, et si la valeur adversative relève de l ’analyse proposée
dans l’exemple précédent.

Denniston classe en ‘(3) Logical (iii)’ (1954: 563), c’est-à-dire «adversative in tone»,
le passage suivant:

κ α ίτ ο ι εγωγε οΐμαι, <5 βέλτιστε, καί την λύραν μοι κρείττον είν α ι
άνάρμοστόν τε καί διαφωνέιν, καί χορόν φ χορηγοίην, καί πλείστους
ανθρώπους μή όμολογείν μοι άλλ’ εναντία λεγειν μάλλον ή ενα οντα
έμέ έμαυτφ άσυμφωνον είναι κ αί άναντία λεγειν. (Plat. G r g . 482b)
«[si tu ne fais pas cette démonstration,..., il est impossible, mon cher Calliclès,
que Calliclès vive en accord avec lui-même et ne demeure pas dans une per­
pétuelle dissonance.] Or j ’estime pour ma part, mon cher, que mieux vaudrait
me servir d ’une lyre dissonante et mal accordée, diriger un chœur mal réglé,
ou me trouver en désacord et en opposition avec tout le monde, que de l ’être
avec moi-même tout seul et de me contredire»

Apparemment, Humbert (1954: 415) classe, lui, cet exemple dans les emplois «en
valeur progressive», qu’il définit ainsi: «καίτοι introduit une id é e qui logiquement
s’ajoute à des développements antérieurs — comme une d é m o n s t r a t i o n , une
i n f é r e n c e , une c o n c l u s i o n au moins provisoire ou partielle». Si l’on s’en tient à la
phrase précédente, on a l’impression d ’un raisonnement logique, presque d’un syl­
logisme. Or nous avons vu que notre particule n ’est pas usuelle dans un pur syllo­
gisme, dans ces syllogismes d ’école qui sont indépendants de l’énonciation. C ’est
que καίτοι est une particule fortement dialogale. Il faut donc prendre en compte la
situation. Socrate a une forte exigence de non-contradiction avec soi-même, au nom
de son attachement à la philosophie. Mais ce n ’est pas du tout la préoccupation de
Calliclès, qui continue à ne pas voir qu’il est inadmissible d ’être fluctuant pour plaire.
(H suffit pour s’en convaincre de lire la suite du dialogue; voir par exemple 5 10-513c,
qui se termine par οΰ πάνυ σοι πείθομαι dans la bouche de Calliclès). Il ne l’a pas
dit explicitement, mais cela découle du portrait que Socrate a fait de lui. Socrate s’en
prend à cette philosophie non formulée. C’est probablement le sens qu’il faut
accorder au «adversative in tone» de Denniston, mais nous sommes loin, d’un point
 BERNARD JACQUINOD

de vue pragmatique, de la seule logique. La distinction faite par Denniston entre

adversatif et logique ne me semble pas pertinente.

Denniston classe aussi en ‘(3) (iii)’ (1954: 563), donc toujours dans «adversative in
tone».

Τ ί δέ δή Κέβητι; εφη ό Σωκράτης· δει γάρ και Κέβητα πείθειν.

Ίκανώς, εφη ό Σιμμίας, ώς εγω γε οίμαν κ α ί τ ο ι καρτερώτατος
ανθρώπων έστίν προς το άπιστεΐν τοΐς λόγοις. (Plat. P h d . 77a6-10)
«Mais pour ce qui concerne Cébès?, reprit alors Socrate; car il faut aussi
convaincre Cébès. — Elle [= cette démonstration] lui suffit, dit Simmias,
autant que je puis croire; et pourtant, c’est le plus résistant des hommes pour
ce qui est de la défiance à l ’égard des arguments» (trad. L. Robin).

Denniston donne ici une excellente description, à laquelle je n ’ai prêtée toute l’atten­
tion qu’elle mérite qu’après avoir formulé la proposition que je fais ici. «Cebes’
normal scepticism is contrasted with his openness to conviction on the present occa­
sion : at the same time, the fact that he is convinced is an indication, a f o r t i o r i , that
anybody else must be convinced» (1954: 563). Mais cette remarque de Denniston est
perdue au milieu d ’une subdivision de sa troisième partie ( l o g i c a l ) , alors qu’à mon
avis elle devrait être la clé des valeurs adversatives. De fait, l’idée que Cébès se laisse
convaincre (A) pourrait donner à entendre qu’il est facile à persuader (c). La proposi­
tion introduite par καίτοι ( B ) vise à imposer la déduction contraire (-c), (ici, B = -c).
On pourrait toutefois reprocher à Denniston de n ’avoir pas dit ou pas vu que ce qui
est corrigé par καίτοι le plus souvent est implicite, et non évident a priori, n’étant pas
le contraire de ce qui précède.

Dans le premier passage que nous avons cité, Plat. P h d . 68d9-12, le début des
paroles de Socrate se déduisait logiquement de ce qui précède: «La crainte de maux
plus grands ne détermine-t-elle pas ceux d ’entre nous qui ont du courage à affronter
la mort, quand il y a lieu de l’affronter?». La proposition introduite par κα ίτοι
apporte un éclairage nouveau à cette réflexion et conduit à la concevoir de façon dif­
férente. Certes, cette nouvelle conception est opposée à la précédente, mais cette
valeur d’opposition réside ici dans le contenu des propositions, et elle paraît forte à
cause de άρα dans la première et de άλογον dans la seconde. Je pense que Dennis­
ton a tort de rejeter l ’idée que la valeur adversative est dans le contenu des proposi­
tions. D ’autre part, pour reprendre ma proposition, on pourrait peut-être aller plus
loin et dire que de la première idée: ‘tous le font’, on pourrait déduire ‘donc c’est
naturel’, et c’est ce présupposé possible qui est battu en brèche, et non le fait indis­
cutable qu’on fait des choses courageuses parce qu’on craint un danger plus grand.

Après cet exemple nous avons cité Plat. C r a t . 401el-3: ici, Socrate dit d’abord (avec
sans doute un peu d ’humour) que pour des raisons plutôt théologiques, il doit, après
 SUR LE RÔLE PRAGMATIQUE DE ΚΛΙΤΟΙ 139

Hestia qui est invoquée en premier dans les sacrifices, traiter de l’étymologie de Rhéa
et de Kronos (il joue aussi sur la locution άφ' Εστίας άρχεσθαι (par ex. E u t h y p h r .
3a). Mais il s’avise qu’il a déjà parlé de ce même Kronos précédemment. On peut
avoir des positions divergentes sur le caractère adversatif ou non; ce qui est sûr, c ’est
que cette remarque, qui ne se tire pas de façon évidente du contexte proche, devrait
modifier le déroulement de l’exposé. Elle ne contredit pas qu’en droit ces deux
divinités doivent être traitées après Hestia (donc, de ce point de vue elle n ’est pas
adversative), mais d ’un autre côté, après avoir dit qu’il faut traiter ces deux mots,
Socrate donne une raison de ne pas le faire (de ce point de vue elle est adversative).
D ’un point de vue argumentatif, les choses sont simples. A une considération qui
oriente la discussion dans un certain sens est ajoutée une autre qui en modifie le
cours. On peut ici encore faire la même description: de A , on conclut «donc nous
allons examiner ces mots», et B suggère, au moins provisoirement, la solution con­
traire. Nous y reviendrons. Nous voyons là que καίτοι est normalement à l’intérieur
d’un débat, mais qu’il suppose une sorte de dédoublement de l ’interlocuteur. Il y a
débat, activité de la pensée, et il faut toutefois relativiser l ’affirmation de Denniston
(cf. page 132): le ton n ’est pas nécessairement passionné, loin de là. Il serait plutôt
enjoué.

D ’ailleurs, καίτοι se trouve presque toujours à l’intérieur des paroles ou de l’exposé

d ’un personnage (ou de l ’auteur). Il lui arrive toutefois d ’ouvrir un «discours»,
comme en

Τοΰτ' έκεΐνο, ώ Φαιδρέ, ούκέτ' αν τό πέρα άκούσαις έμοΰ λέγοντος,

άλλ’ ήδη σοι τέλος έχέτω ό λόγος.
ΦΑΙ. Κ α ίτ ο ι φμην γ ε μεσοΰν αυτόν, καί έρειν τα ίσα π ερί του μή
έρώντος, ώς δέί έκείνφ χαρίζεσθαι μάλλον, λόγων όσα αΰ έχει αγαθά'
(Plat. P h d r . 241d4)
«SOC. — Ça y est, Phèdre. Tu ne pourras entendre un mot de plus de ma
bouche. Tiens plutôt désormais mon discours pour fini.
PH. — Je croyais pourtant que tu en étais à la moitié et que tu l ’équilibrerais
avec un développement sur celui qui n ’aime pas, sur l ’obligation de lui
accorder par préférence ses faveurs».

La différence n’est qu’apparente et d ’ailleurs Denniston présente ce type d’emploi

comme attendu et cohérent avec l’emploi en discours suivi. En effet, le contenu est
celui-ci:
- tu dis, Socrate, que ton discours est fini
- or je pensais qu’après la thèse viendrait l’antithèse, qu’après le περί του έρώντος,
il y aurait un περί τοΰ μή έρώντος.
H y a bien ici un apport nouveau destiné à modifier la pensée de l ’autre et l ’amener à
voir les choses autrement, à modifier son jugement. Phèdre rappelle un principe de
composition rhétorique, que Socrate semble avoir négligé. Ce rappel est destiné à
 BERNARD JACQUINOD

amener Socrate à un prolongement de ce discours qu’il considère comme achevé, ou,

à tout le moins, à une explication sur cet arrêt jugé prématuré. En effet, les paroles
mêmes de Socrate en 238el-2 faisaient présager un discours en deux parties (άπό τε
έρώντος καν μή). Phèdre est donc fondé à attendre une seconde partie, et il exprime
des raisons que Socrate ne prenait pas explicitement en compte. En déclarant le dis­
cours fini, Socrate donnait à penser qu’il était complet, et c’est à cette implication que
s’en prend la phrase introduite par καίτοι.

Καίτον s’emploie aussi dans des parenthèses. Denniston analyse l ’exemple suivant:

ΦΑΙ. Σκώπτεις τον λόγον ημών, ώ Σώκρατες.

ΣΩ. Τούτον μ έν τοίνυν, ΐνα μή συ άχθη, έάσωμεν— κ α ίτο ι συχνά γε
έχενν μον οοκέΐ παραδείγματα προς ά τις βλέπων όνίναιτ’ αν, μιμείσθαι
αύτά επιχειρών μή πάνυ τι—εις δ έ τους ετέρους λόγους ίωμεν. (Plat.
P h d r . 264e3-7)
«PH. — Tu railles notre discours, Socrate ...
SOC. — Eh bien! pour que, toi, tu n’aies pas de chagrin, laissons-le donc en
paix, ce discours! Ce n’est pas pourtant qu’à mon avis il n ’abonde en exem­
ples sur lesquels il y aurait profit à fixer les yeux, en essayant de ne les guère
imiter ! Venons-en plutôt aux discours qui ont suivi».

Ce type d ’emploi semble contredire la définition que j ’ai tenté de donner pour
καίτοι. En effet, Socrate ne tient dans sa conduite aucun compte de l ’objection qu’il
s’est faite à lui-même, et il ne laisse aucune place à une utilisation de cette parenthèse
insérée, comme le remarque Denniston, entre deux phrases dont l ’opposition est
soulignée par le jeu de μέν . . . δ έ (1954: 558), je dirais plus précisément entre deux
impératifs. Or non seulement ce καίτοι, comme les autres qui ont été analysés,
apporte une information nouvelle destinée à faire réfléchir, mais encore il introduit
une considération qui s’en prend au présupposé de ce qui précède. Proposer de ne
plus s’intéresser à un discours pourrait être interprété comme un mépris de son con­
tenu; c ’est une attaque à ce sous-entendu qui est introduite par καίτοι. Avec καίτοι,
on espère un changement dans l’opinion qui pourrait naître chez l’interlocuteur, c’est
la déduction que l’auditeur pourrait faire de l ’affirmation qui est niée, et non l’affir­
mation elle-même. Simplement, Socrate laisse ensuite à son interlocuteur le soin de
modifier sa pensée.

On peut faire la même analyse de

ΑΛ.. Πολύ γέ μοι, <SΣώκρατες, νυν άτοπώτερος αν φαίνη, επειδή ήρξω

λέγειν, ή ότε σιγών ε'ίπου· κ α ίτ ο ι σφόδρα γ ε ήσθ' ίο έίν καί τότε
τοιοΰτος.
(Plat. A le . 1 106a3)
 SUR LE RÔLE PRAGMATIQUE DE KAITOI 141

«Aie.: Vraiment, Socrate, depuis que tu as commencé à parler, je te trouve

maintenant beaucoup plus étrange encore que dans le temps où tu me suivais
sans mot dire! Et c’était bien cependant tout à fait sous ce jour que tu te mon­
trais alors».

Le comparatif «plus étrange» et l’adverbe qui l’accompagne (νΰν άτοπώτερος)

installent en principe l’auditeur dans une échelle argumentative (au sens de O.
Ducrot) qui prévoit que, auparavant, il n ’était pas très étrange. Καίτοι σφοδρά γε
ήσθ' Ιδείν καί τότε τοιοΰτος est destiné à éloigner l ’interlocuteur de cette conclu­
sion naturelle. Il ne s’agit que d’une pointe d ’Alcibiade, sans conséquence sur le
déroulement ultérieur de l’échange, mais elle modifie au passage une idée, en agissant
sur l’implicite, en déviant la pensée de la conclusion qui semblait acquise, ce qui est
bien le rôle général de καίτοι. Bien des emplois de καίτοι gagneraient à être décrits
avec les classes argumentatives, voire avec les échelles argumentatives d ’O. Ducrot.
On voit de façon évidente que le même type d’analyse s’applique parfaitement à
l’exemple qui vient ensuite, avec un Socrate qui plaisante quelque peu, ce qui rend le
passage vivant:

ΑΛ. Τί δέ; ούχ ικανοί διδάξαι ουτοι;

ΣΩ. Οϋκουν τά πεττευτικά γ ε κ α ί τα μιγ κα ίτο ι φαυλότερα αυτά όίμαι
των δικαίων είναι. (Plat. A l e . I 110e4-6)
[Socrate a ironisé sur la réponse d’Alcibiade qui prétend avoir appris du
grand nombre.]
«Alcibiade. — Eh quoi ? le grand nombre n’est-ilpas capable d’enseigner?
Socrate. — Pas même à jouer au trictrac, en tout cas. Et pourtant c’est là une
matière moins délicate que la justice» (trad. Croiset).

3. κα ίτοι et les autres particules

Π serait utile aussi d ’opposer notre καίτοι à d ’autres particules, sans qu’il soit pos­
sible de voir tout le système des particules. Notre καίτοι a des emplois où αλλά
semblerait pouvoir figurer, Mais άλλο: marque une rupture, ce qui n’est pas le cas de
κ α ίτο ι. La particule άλλα peut introduire un contradictoire, καίτοι tout au plus un
contraire et encore n’a-t-on le plus souvent non pas le contraire de ce qui est dit, mais
de ce que pourrait laisser supposer ce qui est dit. On voit bien la différence en

Ουτος γάρ έμοί φαίνεται τά έναντία λέγειν αυτός έαυτω έν τη γραφή

ώσπερ αν ε ί ε’ίπον “’Αδικεί Σωκράτης Θεούς οΰ νομίζων, ά λλα θεούς
νομίζων.” κ α ίτο ι τοΰτό έστι παίζοντος. (Plat. Α ρ . 27a6)
«Car il est clair pour moi qu’il se contredit à plaisir dans sa plainte, qui, en
somme, revient à ceci: ‘Socrate est coupable de ne pas croire aux dieux, bien
que d’ailleurs il croie aux dieux.’ N ’est-ce pas là une simple plaisanterie?»
(trad. M. Croiset)
 BERNARD JA CQUINOD

ou αλλά oppose la même expression au positif et au négatif, tandis que καίτοι intro­
duit un jugement sur la pensée de Mélétos. Cet emploi de άλλα correspond à l’em­
ploi décrit par C.J. Ruijgh: ούκ A άλλα B et appelé «éliminatif» (1971: 127 et 135)
Les deux particules ne sont absolument pas interchangeables dans cet exemple et seul
άλλα est vraiment adversatïf. Toutefois ce n’est pas parce qu’à la suite de καίτοι est
ajouté un jugement qu’il faut réduire notre particule à un rôle additif. Certes, il n’y a
pas d’opposition au contenu de ce qui vient d ’être dit, mais il y a un présupposé, à
savoir qu’une accusation formulée selon les règles du droit devant un tribunal
athénien (έν τη γραφή) est une chose sérieuse, et c’est ce présupposé entièrement
implicite qui va être battu en brèche. On peut rendre compte d ’un tel emploi avec les
échelles argumentatives de O. Ducrot, ou en termes de stratégie discursive.

On peut poursuivre par Γ analyse de:

ΕΤΘ. Ώ ς οΰν καινοτομοΰντός σου περί τα θεία γέγραπται ταύτην την

γραφήν καί ώς διαβαλών δή έρχεται εις τό δικαστήριον είδώς οτι εΰδι-
άβολα τά τοιαΰτα προς τούς πολλούς, κα ί έμοΟ γάρ τοι, όταν τι λέγω έν
τη έκκλησίς: π ερί των θείων, προλέγων αύτοίς τά μέλλοντα, κατα-
γελώσιν ώς μαινομένου- κ α ίτ ο ι ούδέν ότι ούκ αληθές είρηκα ών
προείπον, ά λλ' όμως φθονοϋσιν ήμίν πάσι τοίς τοιούτοις. ά λλ' ούδέν
αυτών χρή φροντίζειν, άλλ’ όμόσε ίέναι.
ΣΩ. ΤΩ φίλε Εύθύφρον, ά λ λ α τό μέν καταγελασθήναι ίσως ούδέν
πράγμα.
(Plat. E u i h y p h r o 3bl0-c7)
«Euth. — ... E déduit de là que tu introduis de nouvelles croyances, et c’est la
raison de sa plainte. Oui, voilà pourquoi il vient te calomnier devant le tri­
bunal; il sait combien cette matière se prête à la calomnie auprès de la foule.
M o i- m ê m e , lorsque je parle de choses religieuses dans l’assemblée, lorsque je
leur prédis ce qui doit arriver, ils me tiennent pour fou et se rient de moi. E t
p o u r t a n t , pas un mot de mes prédictions qui ne soit vrai. Mais, vois-tu, ils
sont jaloux des gens de notre sorte. Q u ' i m p o r t e ! Ne nous soucions pas
d ’eux et osons leur tenir tête.
Socr. — Ah! s’il ne s’agissait que de prêter à rire, mon cher Euthyphron, ce
ne seraitrien. [Nos Athéniens, si je ne me trompe, ne s’embarrassent guère de
l ’habileté qu’ils attribuent à tel ou tel, pourvu qu’il enseigne ce qu’il sait]»,
(trad. M. Croiset)

Le passage présente un καίτοι précédé d’un καί ... γάρ τοι et suivi de trois άλλά.
καί ... γάρ τοι, quelle que soit la façon dont on le segmente, introduit un parallèle qui
ne modifiera pas le cours du raisonnement. A l ’intérieur de ce parallèle, καίτοι a une
valeur concessive «ils se moquent de moi, et pourtant je dis vrai». On s’attendrait à ce
que la moquerie s’appuye sur le ridicule des propos d ’Euthyphron, et l ’échelle
argumentative de la moquerie contient l ’idée d’erreur chez la personne dont on se
 SUR LC RÔLE PRAGMATIQUE DE ΚΛΙΤΟΙ 143

moque. Le nouvel apport est destiné à modifier la conclusion. Et c’est un Euthyphron

désabusé qui propose ensuite un changement d’attitude et qui se résigne à ne pouvoir
discuter avec ses adversaires, qui propose de ne pas se soucier d’eux, et au contraire
de leur tenir tête.

Nous avons déjà parlé de Plat. P h d . 68dll-12 (p. 133), mais le passage contient en
réalité trois καίτοι (un καίτοι et deux κ α ίτο ι... γε) et deux άλλ’ όμως:

ΣΩ. — Τφ δεδιέναι άρα καί δέει άνδρείοί είσι πάντες πλήν οί φιλό­
σοφοι- κ α ί τ ο ι αλογόν γ ε δ έει u v à κ αι δειλίμ άνδρέιον εΐναι.
ΣΙ. — Πάνυ μέν οΰν.
ΣΩ. — Τί δέ οί κόσμιοι αύτών; ού ταύτόν τούτο πεπόνθασιν- ακολα­
σία τινι σώφρονές είσιν; κ α ί τ ο ι φαμέν γ ε αδύνατον είναι, άλλ’ όμως
αύτοις συμβαίνει τούτω ομοιον τό πάθος τό περί ταύττιν την εύήθη
σωφροσύνην- φοβούμενοι γάρ ετέρω ν ηδονών στερηθήναι κ α ί
έπιθυμοΰντες έκείνω ν, άλλων άπέχονται ΰπ’ άλλων κρατούμενοι,
κ α ί τ ο ι καλοΰσί γ ε ακολασίαν τό ύπό των ηδονών άρχεσθαι, ά λ λ '
όμως συμβαίνει αύτοις κρατουμένοις ΰφ' ήδονών κρατείν άλλων ηδο­
νών. (Plat. P h d . 68dl0-69a2)
«Socrate — Ainsi, sauf chez les philosophes, c’est la crainte et la peur qui
rend tous les hommes courageux. Et pourtant, n ’est-il pas illogique d ’être
courageux par peur et lâcheté?
Simtnias — Tout à fait.
Socrate — Passons à ceux d’entre eux qui ont de la tempérance. Ne leur
arrive-t-il pas d’être tempérants par une sorte d’incontinence? Pourtant nous
disons que c’est impossible, mais pourtant cette sotte tempérance ne les en
met pas moins dans une situation qui ressemble à cela: redoutant en effet
d ’être privés de certains plaisirs, et avides de ces plaisirs mêmes, ils
s’abstiennent d’autres plaisirs, alors que d’autres plaisirs les dominent. On
peut bien appeler incontinence la sujétion à l’égard des plaisirs, il n ’en arrive
pas moins à ces gens qui sont dominés par des plaisirs de dominer cependant
d’autres plaisirs» (trad. Jacquinod)

Nous avons vu que le premier καίτοι détruit la conclusion «c’est naturel», que l’on
pourrait tirer de la proposition précédente. De même, le second annule la conclusion
que l’on pourrait tirer de άκολασίςι τινι σώφρονές είσιν et qui serait «c’est possi­
ble, puisque c’est un fait». Avec άλλ’ όμως, Socrate contredit directement son affir­
mation précédente: c ’est impossible ..., άλλ’ όμως, ça arrive. C ’est ce même schéma
qui se reproduit plus loin (un peu moins nettement) = c’est ακολασία qui est nié par
κρατειν άλλων ήδονών, la concession porte plus directement sur ce qui précède, et
il s’agit donc davantage d’une contradiction.
 BERNARD JACQUINOD

D ’autres particules servent pour introduire ia seconde prémisse d’un syllogisme chez
Platon, δέ γε et άλλα μήν; comparons deux exemples du C h a r m i d e , l ’un avec
καίτοι, et l ’autre avec δέ γε:

ΣΩ. Ε νίοτε άρα, ην δ' έγώ, ώφελίμως πράξας ή βλαβερώς ό Ιατρός ού

γιγνώσκει εαυτόν ώς επραξεν- κ α ίτ ο ι ώφελίμως πράξας, ώς ό σός
λόγος, σωφρόνως επραξεν. ή οΰχ ούτως έλεγες;
KP. ’Έγωγε.
ΣΩ. Ο ό κ ο ύ ν , ώς εοικεν, έν ίο τε ώφελίμως πράξας πράττει μεν
σωφρόνως κ α ί σωφρονεί, άγνοει δ’ εαυτόν cm σωφρονεί; (Plat.
C h r m .l6 4 b ll-c 6 )
«Socrate — Ainsi, repris-je, le médecin, que son remède réussisse ou non,
peut avoir agi parfois sans savoir ce qu’il faisait? Cependant, s’il réussit, tu
l’appelles sage. N ’est-ce point ce que tu disais?
Critias — Oui.
Socrate — Par conséquent, si je ne me trompe, quand il guérit son malade, il
agit sagement et il est sage, mais sans savoir qu’il l ’est?» (trad. Croiset)

et
ΣΩ. Φ αίνεται αρα ήμΐν, έφην έγώ, κατά γε τό σώμα οΰ τό ήσύχιον,
α λλά τό τάχιστον κ αί όξύτατον κάλλιστον ον. ΤΗ γάρ;
ΞΑ. Πάνυ γε.
ΣΩ. Ή δέ γ ε σωφροσύνη καλόν τι ην;
ΞΑ. Ναί.
ΣΩ. Οΰ τ ο ίν υ ν κατά γε τό σώμα ή ήσυχιότης άν, ά λ λ ’ ή ταχύτης
σωφρονέστερον ε'ίη, έπειδή καλόν ή σωφροσύνη. (Plat. C h r m . 159d3-ll)
«Socrate — Il est donc évident que, dans les choses du corps, ce n’est pas la
lenteur, mais au contraire la rapidité qui est surtout belle et louable. N’est-il
pas vrai?
Charmide — Tout à fait.
Socrate — Or, c ’est entendu, la sagesse est une belle chose?
Charmide — Oui.
Socrate — Donc, en ce qui concerne le corps, c’est la rapidité, et non la
lenteur, qui est sage, puisque la sagesse est belle»
Noter le rôle de τοίνυν en position conclusive.

Dans le premier passage— décrit par Denniston (p. 562) comme un syllogisme avec
conclusion introduite par ούκοΰν—l ’apport introduit par κ α ίτο ι (ώφελίμως
πράξας, ώς ό σός λόγος σωφρόνως επραξεν) se rattache à l’évidence au même
champ argumentatif que ce qui précède (ώφελίμως πράξας ή βλαβερώς ό ιατρός ού
γιγνώσκει εαυτόν ώς επραξεν), comme en témoigne la reprise des mots, mais
l ’orientation des propositions n ’est pas du tout la même, elles sont presque inverses.
Dans le second, malgré καλόν qui fait écho à κάλλιστον, l’apparition de la σωφρο-
 SUR LE RÔLE PRAGMATIQUE DE ΚΑ11ΌΙ 145

σύνη (argument B ) n’est en rien impliquée par ce qui précède, B est simplement
présenté comme un nouvel élément ayant la même propriété que A . On pourrait for­
maliser ainsi:
Dans le premier exemple, on aurait (en négligeant les conditions d ’application pour
ne pas alourdir les formules):

A καίτοι B, tels que

A =* c {λέγεις o u ό ιατρός * σώφρων}
B = - c = {λέγεις οτι ό Ιατρός = σώφρων}

Dans le second exemple, et sans discuter ce qu’a d’étrange le raisonnement, on aurait

A δ έγ ε B, tels que
étant donné E = {x; x καλόν έστι},
As E
Be E

Le rapport entre les éléments de ces deux «syllogismes» est fort différent et c’est cela
qui justifie le changement de particules; il ne suffit pas donc de parler de syllogisme,
il faut être attentif à la portée de chaque argument par rapport aux autres.
On notera l ’emploi de αλλά pour introduire un contradictoire à τό ήσύχιον, puis à
ήσυχιότης.

L ’emploi de αλλά μην dans un syllogisme se trouve par exemple en Plat. Resp.
464a-b (Denniston, p. 346):

— TAp' οΰν τούτων αιτία πρός τη άλλη καταστάσει ή των γυναικών τε

καί παίδων κοινωνία τοΐς φύλαξιν;
— Πολύ μεν ούν μάλιστα, εφη.
— 'Αλλά μην μέγιστόν γε πόλει αύτό ώμολογήσαμεν άγαθόν, άπεικά-
ζοντες εύ οικουμένην πόλιν σώματι πρός μέρος αύτοΰ λύπης τε πέρι καί
ήδονης ώς έχει.
— Καί όρθώς γ', εφη, ώμολογήσαμεν. (Plat. Resp. 464a-b)
«— Or à quoi attribuer cet effet, sinon à notre constitution en général, mais
plus particulièrement à la communauté des femmes et des enfants entre nos
gardiens?
— Rien de plus certain, dit-il.
— Mais nous avons reconnu que cette unité de sentiment était le plus grand
bien de l’Etat, quand nous avons comparé un Etat bien constitué à un corps
qui partage la douleur ou le plaisir d ’une de ses parties.
— Et nous avons eu raison, dit-il, de le reconnaître.» (trad. Croiset)
 BERNARD JACQUINOD

Ici, άλλα μην introduit une autre réflexion sur cette communauté des plaisirs et des
peines, qui va aboutir à une conclusion sans idée restrictive dans le raisonnement,
sans renversement de perspective.

J’ai tout à fait conscience du caractère provisoire de mes propositions sur καίτοι, car
j ’ai manqué du temps nécessaire pour étudier tous les emplois. A fortiori n’ai-je pas
eu le temps d’explorer de façon différentielle la valeur des particules qui peuvent
paraître proches. En tout premier figure μέντοι, qu’Humbert donne comme un con­
current qui aurait pu faire obstacle à l ’extension de notre καί rot. Sans avoir la pré­
tention de donner un avis sur cette autre particule, je voudrais revenir sur (cf. p. 134)

μετά 8' Εστίαν δίκαιον 'Ρέαν καί Κρόνον έπισκέψασθαι. κ α ί τ ο ι τό γ ε

του Κρόνου όνομα ήδη διήλθομεν. 'Ίσως μ έ ν τ ο ι ούδέν λέγω. (Plat. C r a t.
401el-3)
«Après Hestia, il est juste d’examiner Rhéa et Kronos. Le nom de Kronos, à
vrai dire, nous l ’avons déjà passé en revue. Mais peut-être mes propos sont-
ils sans valeur.» (trad. L. Méridier)

où un καίτοι est immédiatement suivi d’un μέντοι. Cette dernière particule joue ici
un rôle très comparable à celui de καίτοι, car elle va conduire à une inversion de la
pensée. Il y a donc une zone de recoupement entre les emplois de καίτοι et de
μέντοι, dont le rôle dans d ’autres emplois est assez différent. Μέντοι introduit une
idée nouvelle moins prévisible, souvent une surprise, comme c ’est le cas ici. Mais
d ’autres emplois de cette particule sont très éloignés de καίτοι (cf. Plat. R e s p . 583c
Είναι μέντοι).

4. κ α ί τ ο ι chez quelques a u tre s au te u rs

Après cette analyse chez Platon, il est indispensable de jeter un regard en arrière sur
les œuvres antérieures. La plus ancienne attestation serait dans Sappho, mais le frag­
ment nous est parvenu dans un tel état qu’il n ’est pas possible de faire la moindre
analyse. Aucune attestation (sûre) chez Pindare.
Dans ce qui nous reste d ’Eschyle, κα ίτο ι est presqu’absent en dehors de
Prométhée, seule pièce conservée entièrement qui présente καίτοι. Le seul exemple
notable par rapport à ce qui a été dit est au vers 101, καίτοι τί φημι;, qui paraît intro­
duire une rupture: Prométhée, après s’être interrogé sur l’avenir, change de point de
vue et rappelle son omniscience du futur. Il faut voir de près quel est le contexte.
Dans un long monologue, Prométhée parle de ses souffrances et il vient de préciser la
raison de ses sanglots par une question indirecte: quand ses épreuves prendront-elles
fin? Καίτοι introduit une réflexion qui devrait le conduire à se taire; mais ce qu’in­
troduit καίτοι ne contredit pas qu’il y ait une raison pour se lamenter sur la durée de
ses épreuves, et corrige seulement l’idée qu’il pose cette question parce qu’il ne sait
pas. Il sait, et il ne devrait pas avoir l ’air de ne pas savoir. Il faut comprendre ce
 SUR LE RÔLE PRAGMATIQUE DE KAITOI 147

καίτοι τί φημι; comme «pourquoi dire cela?» et non comme «Mais que dis-je?»
(Mazon). Dans ce type d ’emploi, l’objection est explicitée dans la phrase suivante,
mais le rôle de καίτοι reste le même (voir aussi Th. 3.62.3 et Soph. O C 1132). La
particule κα ίτοι annule une fausse interprétation, et annonce un rappel qui doit
modifier et l ’idée de l’auditeur, et aussi le comportement de Prométhée. Ce qui est
proposé pour décrire καίτοι rend bien compte du corpus eschyléen. Καίτοι marque
ici un dédoublement fictif: Prométhée prend soudain conscience que les implications
de ce qu’il vient de dire sont inacceptables, car en contradiction avec son omni­
science. Son esprit s’est égaré, il annonce par ce καίτοι qu’il se ressaisit et qu’il
entrevoit des arguments qui vont à l’encontre de ce qui précède.

Aristophane fait de καίτοι un emploi qui est proche de celui de Platon. On notera
que le plus fort contingent de καίτοι se trouve dans les N u é e s , où cette particule se
trouve 11 fois, alors qu’en moyenne elle n ’apparaît que 4 fois par pièce, et, surtout on
la rencontre 4 fois du vers 1045 au vers 1082. Or cette pièce met en scène le raison­
neur par excellence, Socrate, et les vers 1045-1082 sont dans le débat entre le
Δίκαιος Λόγος et Γ "Αδικος Λόγος. C’est une pièce dans laquelle on raisonne
beaucoup. Cela confirme, s’il en était besoin, la place de καίτοι dans l ’argumenta­
tion.

Sophocle présente douze emplois de καίτοι, dont sept de κ α ίτ ο ι.... γε. Presque
tous correspondent bien à la valeur adversative telle que je l’ai décrite.
Un seul fait difficulté, c’est P h il. 1257, parce que, là, καίτοι est au début des
paroles d’Ulysse:

Κ α ίτ ο ι σ’ έάσω· τφ δέ σύμπαντι στρατφ

λέξω τάδ' έλθών, ος σε τιμωρήσαντι (Soph. P h i l . 1257-8)
«Eh bien! non; malgré tout, je préfère te laisser là. C’est à l’armée entière que
j ’irai faire mon rapport, et c’est elle qui te châtiera.» (trad. Mazon)

Néoptolème refuse de céder à Ulysse, qui menace d’employer la force (1254-5 «tu
vois ma main: elle se porte à la garde de mon épée»). Néoptolème menace de
dégainer lui aussi (1255-6), aussi Ulysse change-t-il d’attitude, ne se chargeant plus
lui-même d’accomplir sa menace. Il y a un changement complet, un vrai revirement, et
on attendrait dans ces conditions une autre particule. Je suggère de justifier ici l’em­
ploi de καίτοι par une volonté de la part d ’Ulysse de masquer une reculade un peu
honteuse en lui donnant l ’aspect d ’une simple modification, d’un changement
raisonnable, en employant la particule par excellence de l ’apport nouveau dans le
raisonnement.
 BERNARD JA CQU1N0D

5. R eto u r à P laton et conclusion

Il est temps d’étudier le seul exemple un peu particulier de Platon, en I o n 533c, pas­
sage dans lequel καίτοι introduit un impératif: καίτοι ορα τούτο τί εστιν

ΙΩΝ Ούκ έχω σοι περί τούτου άντιλέγειν, ώ Σώκρατες· άλλ’ έκεινο
έμαυτω σύνοιδα, οτι περί 'Ομήρου κάλλιστ' ανθρώπων λέγω και ευπορώ
καί οί άλλοι πάντες μέ φασιν εύ λέγειν, περί δέ των άλλων ου. κ α ίτο ι
ορα τούτο τί εστιν.
ΣΩ. Καί όρώ, ώ 'Ίων, καί έρχομαι γέ σοι άποφαινόμενος ο μοι δοκεί
τούτο είναι. (Plat. I o n 533c)
«Ion. — Je ne puis te contredire là-dessus, Socrate; mais j ’ai conscience que
sur Homère je parle mieux que personne, j ’abonde en idées, et tout le monde
reconnaît mon talent de parole, tandis que pour les autres il n ’en est rien. Vois
pourtant ce que cela signifie.
Socrate. — Je le vois, Ion, et même je m’en vais te faire connaître ce que cela
signifie, selon moi» (trad. Méridier)

L ’emploi de καίτοι malgré l’impératif me semble pouvoir s’éclairer de la même

façon que tous les autres emplois platoniciens. Ion vient de reconnaître ses limites,
par opposition à ceux dont Socrate vient de parler. On pourrait en déduire que son
cas ne mérite pas une réflexion, d’être un sujet de débat. C’est pour que Socrate ne
se laisse pas aller à une telle conclusion que Ion emploie ici un καίτοι.

Pour résumer, καίτοι est rapidement devenu une particule servant essentiellement
dans l’argumentation. Cette fonction argumentative fait qu’il est plus fréquent dans la
prose que dans la poésie, théâtre excepté, le théâtre étant souvent le lieu d’un affron­
tement rhétorique, voire dialectique. C’est un marqueur dont le rôle est de souligner
l ’introduction d’une idée qui doit faire réfléchir sur ce qui vient d ’être dit. Sa fonc­
tion est éminemment pragmatique. La nouvelle orientation qui est suggérée est sou­
vent opposée à celle que l’on pourrait attendre de ce qui précède, mais ce qui est à
modifier, ce n ’est pas ce qui vient d ’être dit, mais ce que l ’on pourrait conclure de ce
qui vient d’être dit. Dans son emploi le plus caractéristique, καίτοι indique donc
qu’il faut renoncer à une conclusion que l ’on pourrait pouvoir déduire de ce qui
précède. C’est son rôle principal chez Platon comme chez Sophocle, son rôle majori­
taire chez Hérodote. Il faudrait bien sûr distinguer les auteurs. Chez Thucydide,
κα ίτο ι introduit le plus souvent un rappel d ’un principe qui condamne ce qui
précède. Il faudrait aussi tenir compte de l’étude de Blomqvist sur l’époque hellénis­
tique. Ce dernier reprend le plan de Denniston, ce qui brouille les choses, mais il
faudrait sans doute admettre un affaiblissement du sens de καίτοι, probablement dès
Aristote.
Pour conclure, il me semble que καίτοι a comblé un vide: il n’y avait pas, si je lis
bien A u t o u r d e τ ε é p i q u e de particule consistante (non monosyllabique) spécialisée
 SUR LE RÔLE PRAGMATIQUE DE ΚΛΙΤΟΙ 149

dans ce rôle. L ’essor de la pensée liée au développement de la rhétorique et de la

dialectique a rendu nécessaire la spécialisation d’une particule dans ce type d’emploi
concessif. Cette spécialisation s’est faite en complémentarité avec καίπερ qui, bien
qu’étant aussi concessif, l ’est de façon profondément différente d’un point de vue
pragmatique, dans la mesure où elle fait partie de la proposition dont elle restreint la
portée.

Bibliographie

Anscombre, A J.C. et O. Ducrot

1975 ‘L’argumentation dans la langue’. Actes de la session de linguistique de
Saint-Flour. Paris: ENS
Armengaud, F.
1985 La pragmatique. Paris: P.U.F.
Blomqvist, J.
1960 Greek Particles in Hellenistic Prose. Lund
BoHing, G.M.
1902 ‘K a i m with Participle’. AJPh 23,319-321
Denniston, ID .
19542 The Greek Particles. Oxford: Clarendon
Ducrot, O.
1972 Dire et ne pas dire. Paris: Herrmann
1975 La preuve et le dire, Paris
1980 Les échelles argumentatives. Paris: Les Editions de Minuit
1984 Le dire et le dit. Paris: Les Editions de Minuit
Humbert, J.
19542 Syntaxe grecque. Paris: Klincksieck
Kühner, R. und B. Gerth
1898-1904 Ausführliche Grammatik der griechischen Sprache. Satzlehre (2
volumes). Hannovre: Hahnsche Buchhandlung
Récanati, F.
1981 Les énoncés performatifs. Paris: Les Editions de Minuit
Ruijgh, CJ.
1971 Autour de τε épique. Etudes sur la syntaxe grecque. Amsterdam: Hakkert
Schwyzer, E. und A. Debrunner
1950 Griechische Grammatik. Zweiter Band: Syntax und syntaktische Stilistik.
München: Beck
 Η Γ Α Ρ ΊΝ QUESTIONS

A . MARIA VAN ER P TAALMAN KIP

Universiteil van Amsterdam

In this article I shall make a number of observations on the use of η γάρ (sometimes
ή ... γάρ) in questions, not on the elliptical use of this combination (in the sense of
‘isn’t it?’), but on ή γάρ as an introduction to an actual question.1 Tragedy offers a
great many examples of such questions, which share the following characteristics, η
γάρ questions are used when a character is asking for confirmation of something that
has been said, implied or suggested by the previous speaker. As a rule, the tone is
urgent, expressing anxiety, indignation, hope or surprise, depending on the nature of
the information the speaker has just been given. They occur in a stichomythia or a
near-sdehomythia. A number of examples will be discussed.

1. Aeschylus, P r o m e t h e u s 757. Prometheus has just announced that his sufferings

will not end until Zeus will fall from power, whereupon Io asks:

η γάρ ποτ' έστιν έκπεσεΐν αρχής Δία;

Ts it really possible that Zeus will once fall from power?’

She can hardly believe it, but, as she concedes in 759, she would be glad if it hap­
pened, since Zeus is the cause of her terrible distress.

2. Sophocles, O e d i p u s T y r a n n u s 1039. Oedipus asks whether it was his father or his

mother who pierced his ankles, assuming that the messenger from Corinth knows the
identity of his parents. But the messenger does not:

οΰκ οΐδ’· ό δους δέ ταΰτ' έμοΰ λφον φρονεί.

Τ do not know. The man who gave the child to me is better informed about
that than I am.’

1 Denniston (G .P . 284-5) lists these questions without comment, suggesting that the elliptical use
o f ή γάρ; follows naturally from its use in full-blown questions. He docs not note the essential
difference: with ή γάρ questions speakers are asking for confirmation of the words of others, with
elliptical f j γάρ for confirmation of their own words. It is probably for this reason that Kiihncr-
G enh explain elliptical fj γάρ as follows: ή γάρ das in die Frageform umgesetzte versichernde ή
γάρ, wahrlich ja, daher = nicht wahr? ’ (2,337).
As for Plato this article will somewhat diverge from the lecture I gave at the colloquium, since I
was misled by the capriciousncss of Pandora. My case histories were, in the first instance, not
complete. This time they are, but the essence of my argument has not been affected.
 A. MARIA VAN ERP TAALMAN KIP

This information is new to Oedipus; he did not know that it was not the messenger
himself who found him. He eagerly asks for confirmation:

η γάρ παρ' άλλου μ’ έλαβες ούδ' αυτός τυχών;

‘What, thou hadst me from another? Thou didst not light on me thyself?’
(translation Jebb)

3. Sophocles E l e c t r a 1221-1222. When Electra asks where her brother has been
buried, Orestes replies that living people have no tomb. In the next line he assures her
that he is not lying, and Electra, overwhelmed with sudden hope, asks:

η ζ-rj γάρ άνήρ;

‘The man is a l i v e V

And after hearing Orestes’ answer she once more asks:

ή γάρ συ κείνος;
'W h a t l Are you h e V

4. Euripides, H e l e n 784. After being reunited with Menelaus, Helen expresses her
fear that he has only been preserved from death in battle in order to die by the hand
of the Egyptian king. When Menelaus fails to understand how he could possibly
deserve such a fate, Helen explains:

ήκεις αελπτος έμποδών έμοίς γάμοις

‘Your coming is an undesirable obstacle for his marriage with me. ’

On hearing this explanation, Menelaus responds indignantly:

η γάρ γαμείν τις ταμ’ έβουλήθη λέχη;

‘What? Has someone conceived the desire to marry my wife? ’

5. Euripides, I p h i g e n e i a a t A u l i s 325. Agamemnon’s letter, designed to prevent the

arrival of Iphigeneia, has been taken by Menelaus. When Agamemnon asks him to
return it, his brother answers:

οΰ, πριν αν δείξω γε Δαναοίς πασι τάγγεγραμμένα

‘No, not before I have shown the content to all the Greeks.’

These words imply that he has read the letter, thus provoking the angry question:

η γάρ όίσθ’ α μή σε καιρός εΐδέναι, σήμαντρ’ άνείς;

 ΗΓΑΡ IN QUESTIONS 153

‘Did you break the seal? Do you really know what you have no business
knowing?’

So much for my examples from tragedy, only five out of many. Outside tragedy, ή
γάρ questions are comparatively rare. This is not surprising, since they are generally
part of rapid dialogue. I even suspect that they originated in drama and were confined
to the Attic dialect. What i s surprising, however, is that they do not occur in Aristo-
phanic comedy, despite the fact that comedy abounds in dialogues in which ή γάρ
questions would seem at home. Why, then, are they missing? This may have some­
thing to do with ‘register’: probably these questions had a tragic ring to them, sug­
gesting an emotional intensity that was not in keeping with comedy, except in pas­
sages containing tragic parody. Admittedly, they are not found in parody either, but
that may be coincidental.
In prose dialogue η γάρ questions are not completely absent. We find some
examples in the works of Xenophon. In O e c o n o m i c u s 4.20-24 Socrates relates how
Cyrus showed Lysander his ‘paradise’. The guest expresses his admiration for the
designer, whereupon Cyrus tells him that he himself designed the garden and that he
even did some of the planting. Then Lysander asks:

τί λέγεις ... <5 Κύρε; ή γάρ σύ τάίς σάίς χερσί τούτων τ ι έφύτευσας;
‘W h a t do you say, Cyrus? Did y o u really plant part of this with your own
hands?’

It is easy to understand why Xenophon chose ή γάρ here. Lysander, as appears also
from τί λέγεις (cf. πώς έίπας in E l e c t r a 1219) is struck with wonder and asks for
confirmation of an almost incredible statement.
Plato constantly makes use of elliptical η γάρ, but a computer search revealed
only seven occurrences where η γάρ is used as an introduction to a question. Such
questions are clearly too emotional for the calm tone of the philosophical dialogue.
For this reason, it may be interesting to study the passages where they do occur. I
shall start with G o r g i a s 469M0. For some time Polus has been highly irritated, not
to say outraged, by the absurdity of Socrates’ views. Thus he says (467bl0):
σχέτλιά γε λέγεις καί υπερφυή, <3 Σώκρατες, —surely a remarkable choice of
words in a Platonic dialogue. Σχέτλιος occurs only three times in Plato and only
here it is used in connection with words rather than people. And although the adverb
ΰπερφυώς is not exceptional, the adjective is. We may well be sure that Polus is char­
acterised by the language he uses. Thus it is not surprising that when Socrates finally
concludes that committing injustice is the worst thing there is, he indignantly
exclaims:

η γάρ τούτο μέγιστον; ού τό άδικεισθαι μεΐζον;

‘Is t h a t the worst thing? Is suffering injustice not worse?’
 A. MARIA VAN ERP TAALMAN KIP

The particles η γάρ suggest that, like a hero in tragedy, he is experiencing great emo­
tional stress.
Next we turn to G o r g i a s 494e. Callicles has just been confronted with the
inescapable conclusion that, according to his own reasoning, catamites are fortunate.
‘The unshockable Callicles is shocked at la s t...’, as Dodds observes, and thus he
asks indignantly:

ούκ αισχύνη εις τοιαΰτα άγων, ώ Σώκρατες, τούς λόγους;

‘Are you not ashamed, Socrates, to bring the discussion round to such sub­
jects?’

Thereupon Socrates Tetorts:

ή γάρ έγώ άγω ενταύθα, ώ γενναίε, η έκεΐνος ός άν φη άνέδην ουτω

τούς χαίροντας, όπως άν χαίρωσιν, εΰδαίμονας είναι, και μή διορίζηται
των ηδονών όποΐαι άγαθα'ι και κακαί;
Ts it really m e who brings it around in that direction, my noble friend, or is it
the man who says that those who enjoy themselves recklessly, in whatever
way, are fortunate and who does not distinguish between good pleasures and
bad ones?’

With ή γάρ Socrates teasingly suggests that he is sharing the emotional state of his
opponent and that he, too, is filled with indignation. However, an alternative has no
place in an η γάρ question and I think we would do well to mark the difference in
tone by placing a question mark not only after κακαί but also after ώ γενναίε.
In P r o t a g o r a s 33 le it is once again Socrates himself who poses an ή γάρ ques­
tion. This time the tone of his opponent is calm and there is no display of theatrical
emotion. But there is surprise, as Socrates confirms by his own comment:

καί έγώ θαυμάσας έίπον προς αυτόν ή γάρ οϋτω σοι το δίκαιον καί τό
όσιον προς άλληλα έχει, ώστε όμοιόν τι σμικρόν έχειν άλλήλοις;
‘And filled with astonishment, I said to him: “Do you really think that the
relation between the just and the holy is such that they have only some slight
similarity to one another?”’

No such comment appears in M e n o 77c, since this dialogue is written in direct dra­
matic form. But we may well conclude that here, too, Socrates is expressing his
astonishment at the opinion of his opponent:

η γάρ δοκεΐ τίς σοι, ώ Μένων, γιγνώσκων τά κακά ότι κακά έστιν
όμως έπιθυμεΐν αύτών;
‘Do you really think. Meno, that someone, knowing bad things for what they
are, nevertheless yearns for them?’
 Η ΓΑΡ IN QUESTIONS 155

While the above examples meet all the requirements of a tragic ή γάρ question, the
others do not. In Respublica 607c Socrates is delivering his final indictment against
tragedy and against Homer, whom he refers to as the first of the tragedians. He has
just reminded his listeners of the battle between poetry and philosophy, and now he
concludes as follows:

όμως δέ είρήσθω δτι ήμεΐς γε, ε’ί τινα έχοι λόγον είπεΐν ή προς ηδονήν
πονητική καί ή μίμησις, ώς χρή αυτήν είν α ι έν π όλει ευνομούμενη,
άσμενοι αν καταδεχοίμεθα, ώς συνισμέν γε ήμΐν αύτοΐς κηλουμένοις
ύπ’ αυτής· άλλα γάρ τό δοκοΰν αληθές οΰχ δσιον προδιδόναι. ή γάρ, ώ
φίλε, ού κηλή ύπ’ αύτης καί σύ, και μάλιστα όταν δι' Όμηρου θεωρής
αυτήν;
‘Nevertheless, we must say that, if mimetic poetry, that aims at giving us joy,
were able to present good reasons why it must be present in a well-ordered
state, we would welcome it back with pleasure, for we are aware that we our­
selves are bewitched by it. However, it is not allowed by the gods to betray
what we believe to be true. Are not you too, my friend, bewitched by it, and
especially when you behold it through Homer’s eyes?’

Here, everything seems out of place. Socrates is not asking for confirmation of
something just said by another, but of something he himself has just said, and thus
there is no cause for surprise or emotion. Moreover there is the negation ού, which is
surely inappropriate after ή γάρ. Why did Plato choose these particles, which seem
quite uncalled for, while a simple ού would have sufficed? At first I thought it was
meant as a subtle joke. One of the reasons why mimetic poetry must be banished is
its display of feeling. And now, at the very moment when he arrives at his final ver­
dict, Socrates once more reminds his audience, by way of ή γάρ, of these undesirable
tragic emotions. However, if Plato had meant it as a joke, he would have taken care to
use the particles in their proper, iTagic way. As it is, it would perhaps be best to sim­
ply change the punctuation, so that the passage reads: ή γάρ, ώ φίλε; ού κηλή ύπ’
αύτης καί σύ, καί μάλιστα όταν δι’ 'Ομήρου θεωρής αύτην; For this midway
position of elliptical ή γάρ we may compare Theaetetus 160e: ή γάρ, ώ Θεαίτητε;
φώμεν ... κτλ. There, like here, elliptical ή γάρ is followed and explicitated by an
actual question. In Theaetetus Socrates wants to know whether his rendering of
Theaetetus’ opinion was correct, in Respublica whether he was right in saying: ‘we
are aware that we ourselves are bewitched by it.’ In Hippias minor 363c we have a
comparable case. Eudicus assures that Hippias will answer Socrates’ questions and
subsequently asks Hippias for confirmation o f his words; a simple ή γάρ; would
have sufficed, but here again it is explicitated by a full-blown question. Consequently
we should read: ή γάρ, <SΙππία; έάν τι έρωτςί κτλ.
Finally Euthydemus 300b. Ctesippus, by now skilled in wordplay, produces two
more examples:
156 A . MARIA VAN ERP TAALMAN KIP

άλλά μοι δοκείς, Εύθύδημε, où καθεύδων έπικεκοιμήσθαι καί, <εί> οίόν

τε λέγοντα μηδέν λέγειν, καί σύ τούτο ποιέΐν.
‘Me thinks, Euthydemus, that you are asleep while waking, and that, if it is
possible to say n o t h i n g (o r: t o talk n o n s e n s e ) while speaking, you are doing
precisely that.’

At this moment Dionysodorus intervenes in the discussion:

η γάρ οΰχ οΐόν τ', εφη ό Διονυσόδωρος, σιγώντα λέγειν;

‘Is it not possible to speak while being silent’, or ‘to speak of things that are
silent?’

This time the η γάρ question is put by a ‘third man’, but that need not be a decisive
objection. We also find, as in R e s p u b l i c a 607c, the negation on, and although here
one could perhaps defend it by connecting it solely with οΐόν τε (‘is it really impos­
sible...?’), the contents of the question definitely rule out the use of ή γάρ.
Dionysodorus, instead of asking for confirmation, starts an interrogation of his own.
Seizing upon the words ‘to say nothing while speaking’ he inquires about the pos­
sibility of a case that is the reverse (to speak while being silent) or at least very differ­
ent (to speak of things that are silent). In my view, his question is introduced by a
simple ού, while η γάρ again appears to be used elliptically. If this is correct, the par­
ticles do not form part of Dionysodorus’ question, but are rather the conclusion of
the words spoken by Ctesippus: do not you agree, Euthydemus, that you say nothing
while speaking?
I conclude that the whole of Plato’s work yields only four non-elliptical η γάρ
questions: P r o t a g o r a s 31 le, G o r g i a s 469b, 494e, M e r to 77c.
 PARTICLES IN QUESTIONS IN PLATO

C .M .J. SICKING
Rijksuniversiteit Leiden

1. Introduction

When studying Plato’s use of particles in questions1we will have to depart from the
fact that, in Socratic dialogue, the great majority of questions and answers cannot be
described as more or less autonomous moves with their corresponding counter­
moves. We will have to realize that, in typical Socratic conversation, Socrates will
confront his conversation partner with a well-organized series of interconnected
predications—each member of the series being one (and, preferably, only one) step
in a systematic and coherent argument conducted by Socrates in such a way as to
lead up to a final conclusion. So, although a series of questions, of course, is not a
piece of connected speech in the strict sense of the word, the subsequent questions
must be interpreted against the background of the context that gives them their rele­
vance: they must be dealt with as parts of the structured argument in which they
occur.
The separate steps in the argument are, most of them, presented as questions
rather than as statements because Socrates wants his conversation partner to sub­
scribe consciously and of his own free will both to the premisses and to the interme­
diate steps of the argument. If his partner refuses to concur, Socrates’ own belief will
not necessarily be affected: he will threaten to stop the conversation, go back to
something agreed on earlier, make a fresh start or find yet another way to make his
interlocutor see that he has to give up his position. A fair number of the questions
involved, therefore, must be described as assent seeking or ‘apodeictic’ questions
rather than as veritable ‘yes/no questions’, ‘alternative questions’ (πότερον ... ή) or
‘information questions’.
I will argue that the primary function of a number of particles typically used by
Socrates when he is questioning his interlocutors is to enable them to relate each
question to what has preceded and/or to what is to follow, and to trace the articulation
and overall design of the argument. The particles most frequently used for these pur-

1 The following argument is based upon the complete material in the Gorgias. I have used the texts
o f Burnet (Platonis opera, recognovit brevique adnotatione critica instruxit loannes Burnet, III.
Oxford 1903) and Dodds (E.R. Dodds, Pialo, Gorgias. Λ revised Text with introduction and
Commentary. Oxford 1959). Translations have been adapted from W.D. Woodhead (in: Edith
Hamilton & Huntington Cairns [edd.], The Collected Dialogues o f Plato. New York 1961).
158 C.M .J. SICKING

poses are οΰν, ούκοΰν, and (ούκ) άρα. The most frequent of these is ούκοΰν (ca.
90 occurrences in the G o r g ia s alone).

2. ούκοΰν

2.1. When we seek to describe the contribution of ούκοΰν to the utterances in which
it occurs we will have to start from the annoying fact that the accentuation even of our
oldest manuscripts cannot be trusted to represent the author’s intentions: we will
have to make out ourselves whether to write ούκοΰν or ούκοΰν—taking it for
granted that the ancient grammarians were right in saying that, in Greek of the classi­
cal period, ούκοΰν (in which the negation is the predominant element) is to be distin­
guished from ούκοΰν (in which ow is preponderant).
According to Kühner-Gerth, des Places and Denniston, in Plato, ούκοΰν-
s t a t e m e n t s are extremely rare. Denniston (433) is even inclined to insert a question
mark after a l l Platonic instances of ούκοΰν.2 As to ούκουν-questions, Kiihner-Gerth
(2, 166) believe that these are practically restricted to drama, and especially to
Tragedy: «die Frage mit ούκοΰν gehört der aufgeregten und pathetischen Rede an,
die aus einem leidenschaftlichen, unwilligen, erzürnten, erstaunten, ungeduldigen
Gemüte hervorgegangen ist, wie sie besonders in den Tragödien zu sein pflegt.»3 «In
der ruhigen, gemässigten Rede», characteristic of Socratic dialogue, ούκουν-
questions are believed to be out of place. This is in accordance with editorial practice:
in the G o r g ia s there is only one instance of ούκοΰν (out of three)4 where Burnet and
Dodds print a question mark, apparently because they take it that Socrates’ question
can be characterized as ‘impatient’:

(1) ΣΩ. ορα ούν ε ί έθελήσεις έν τφ μέρει διδόναι έλεγχον ά π ο κ ρ ι ν ό μ ε ν ο ς

τά έρωτώμενα. εγώ γάρ δή οιμαι και έμέ καί σε και τούς άλλους

2 There seems to be no direct conclusive evidence for prosodic cues (cf. Devine & Stephens 542-5)
distinguishing between statements and yes/no questions in ancient Greek. Aristotle, Poetics
1456M0, 1457a2l,can hardly be considered to provide such evidence—υποκριτική including mas­
tery both of tone/intonation and gesture. This is, o f course, not to say that, in Greek, there were no
prosodic cues enabling a speaker to characterize an utterance as a yes/no question. But, since these
cannot be observed in a written text, we will have to go mainly by what can be gathered from:
- the presence of an interrogative particle (although identifying these often involves some circular
reasoning);
- characteristic word order (although we know, perhaps, less about Greek word order than seems
desirable);
- the presence of cues in the preceding context, such as: είπέ μοι, and—less reliable—characteris­
tics marking the following context as die answer to a question;
- considerations of content (cf. Gorg. 453e2: ή άριθμητική ού διδάσκει ήμάς οσα έσίιν τά τού
αριθμού, καί ό αριθμητικός άνθρωπος;—where substituting a period for the question mark
would result in something clearly not intended).
3 Cf. also 2, 167: «Durch eine Frage mit ούκοΰν wird nicht, wie durch die mit ούκοΰν, etwas
gefolgert, was sich aus einem anderen Gedanken von selbst ergiebt, sondern es wird ein bestimmtes
und festes Urteil von einem, dessen Gemüt heftig erregt is, mit allem Ernste in der Form einer Frage
ausgesprochen.»
 PARTICLES IN QUESTIONS IN PLATO 159

άνθρώπους το άδικεΐν τοΰ άδικεισθαι κάκιον ήγέϊσθαι καί τό μή διδό-

ναι δίκην τοΰ διδόναι.
ΠΩΛ. Έγώ'οέ γε ούτ- έμέ ούτ' άλλον ανθρώπων ούδένα. έπεί σύ δέξαι'
άν μάλλον άδικεισθαι ή άδικειν;
ΣΩ. Καί σύ γ* άν καί οί άλλοι πάντες.
ΠΩΛ. Πολλοΰ γε δει, άλλ' ούτ' έγώ ούτε σύ ούτ' άλλος οΰδείς.
ΣΩ. Οϋκουν ά π ο κ ρ ι ν η ;
ΠΩΛ. Πάνυ μεν ο ύν καί γάρ επιθυμώ είδέναι ö u ποτ’ έρεΐς. (474b 1-
c3).

Denniston’s interpretation of οϋκουν (432) is the same as that of Kiihner-Gerth, but

he is less certain about its unsuitability for Socratic dialogue: «Broadly speaking this
is no doubt true; but even Socrates is capable, at times, of surprise, and the people
with whom he converses are not invariably ‘ruhig’.» He advocates substituting
οϋκουν for ούκοΰν in a number of passages «where the particle introduces a lively,
suprised, or indignant question». Among his instances are three passages from the
G o r g i a s [(2) - (4)], all of them spoken not by Socrates but by one of his conversa­
tion partners.5 This is in itself not surprising since Denniston will have searched for
examples of lively questions in a dialogue that counts among its d r a m a tis p e r s o n a e
some of the most vehement of Socrates’ interlocutors.

(2) ΣΩ. Ούκοΰν καί περί τάς άλλας άπάσας τέχνας ωσαύτως έχει ό ρήτωρ
καί ή ρητορική· αυτά μεν τά πράγματα ούδέν δει αυτήν είδέναι όπως
έχει, μηχανήν δε u v a πειθοΰς ηύρηκέναι ώστε φαίνεσθαι τοις ούκ
είδόσι μάλλον είδέναι των είδότων.
ΓΟΡ. Οϋκουν [ούκοΰν Burnet, Dodds] πολλή ραστώνη, ώ Σώκρατες,
γίγνεται, μή μαθόντα τάς άλλας τέχνας άλλά μίαν ταύτην, μηδέν έλατ-
τοΰσθαι των δημιουργών; (459b6-c5)
W e l l , i s n ’t th a t a g r e a t c o m f o r t , to b e a b le w i t h o u t le a r n in g a n y o t h e r a r t s
b u t th is o n e to p r o v e in n o w a y i n fe r io r to th e s p e c ia lis ts ?

(3) φημί γάρ, ώ Πώλε, έγώ καί τούς ρήτορας καί τούς τυράννους δ ν ν α σ θ α ι
μέν έν τάίς πόλεσιν σ μ ι κ ρ ό τ α τ ο ν , ώσπερ νυνδή ελεγον- ούδέν γάρ
ποιειν ών βούλονται ώς έπος είπείν, ποιέίν μέντοι ö u άν αύτοΐς δόξη
β έ λ ^ τ ο ν είναι.
ΠΩΛ. Οϋκουν [Dodds; ούκοΰν Burnet] τούτο ëau v τ ο μ έ γ α δ ν ν α σ θ α ι · ,
(466d7-e3)
W e ll, i s n ’t th a t h a v in g g r e a t p o w e r ?

(4) ΣΩ. άλλά καί συ έμοΰ άκουε, ö u άποκτενέί μέν, άν βούληται, άλλά
πονηρός ών καλόν κάγαθόν όντα.

5 447b7 (ούκοΰν ö tav βούλησθε παρ' έμέ ήκειν οίκαδε ) is, of course, another matter: the
speech act involved there is not a question but a courteous invitation.
160 C.M.J. SICKING

ΚΑΛ. Ούκουν [ούκοΰν Bumet, Dodds] τούτο δή καί τό άγανακτητόν;

(511b3-6)
Is not this the very feature thatfills one with indignation?

What the instances (2) - (4) have in common is not primarily that they are ‘lively’,
‘emotional’, ‘indignant’ or ‘surprised’ questions. The rhetorical impact—or illocu­
tionary force—of this type of ‘question’ is equivalent to that of an objection: ‘But
this is precisely what makes rhetoric such a delighful instrument’, ‘But this is pre­
cisely what it means to have great power’ &c. The utterances involved must be
described as rhetorical ‘answer-questions’—the speaker concluding from what has
been said that his conversation partner cannot possibly himself believe what he is
saying: ‘Am I to take it that you intend to deny what seems obvious?’ The question
in (1) is different only in so far as it does not concern Polus’ beliefs but his present
way of acting: ‘Am I to take it that you will not answer my question?’
Characteristic instances of this idiom in drama are:

(5) συ δ’ ούκ άπειρος, ούδ' έμοΰ διδασκάλου

χρήζεις' σεαυτόν σώζ' όπως έπίστασαν
εγώ δέ τήν παρούσαν αντλήσω τύχην,
έστ' αν Διάς φρόνημα λωφήση χόλου.
- ούκουν, Προμηθεύ, τούτο γιγνώσκεις, ότι
οργής νοσούσης είσίν ιατροί λόγοι; (A. Prom. 377-80)

(6) ούκουν δεινόν ταυτί σε λέγειν δήτ' έστ’ έμέ καί διαβάλλειν
προς ’Αθηναίους καί τον δήμον ...; (Ar. E q . 810-11)

(7) ούκουν ταυτί δεινόν άκούειν, ώ Δήμ’, έστίν μ’ ύπό τούτου,

ότιή σε φιλώ; (ib. 820-1)

and also

(8) - ‘I have come to sec how you are.’

- ούκουν (Diggle;6 cett. edd.: ούκοΰν) όρφς μου πρώτον ώς ξηρόν δέμας
(Ε. El. 239).

On the other hand, we may consider printing ούκοΰν in a number of instances out­
side Plato where editors generally have printed ούκουν, such as:

(9) - . . . (‘Isn’t it a shame that I, a god)

αυτός βαδίζω καί πονώ, τούτον δ’ όχώ,
ϊνα μη ταλαιπωροΐτο μηδ’ άχθος φέροι;

6 Denniston (Euripides, Electra. Oxford 1939 ad loc.): «I think that interrogative ούκουν should
probably be substituted here and almost everywhere for interrogative ούκοΰν in drama.»
 PARTICLES IN QUESTIONS IN PLATO 161

- ού γάρ φέρω γώ; - πώς φέρεις γάρ δς γ' όχεΐ;

- φέρων γε ταυτί. - τίνα τρόπον; - βαρέως πάνυ.
- ούκοΰν [VA; Dover: οϋκουν] το βάρος τοΰθ' δ σύ φέρεις δνος φέρει;
- ού δήθ’ δ γ' έχω 'γώ καί φέρω, μά τον Δί’ ού. (Ar. Ran. 23-8)

Here the point apparently is that Xanthias refuses the affirmative answer Dionysus
has been leading up to in an almost Socratic way.

2.2. Outside Plato and Xenophon, ούκοΰν-questions are indeed extremely rare—if,
that is to say, we accept the current punctuation conventions. But this is not because
οΰκοΰν-questions are «preeminently suited to the even, unemotional character of
intellectual discussion», but because they are bound up with Socrates ’ characteristic
way of offering his interlocutors a sequence of interconnected questions, and, in
doing so, making it clear how each subsequent move relates to the demonstration it is
part of. I will limit myself to discussing three characteristic passages.
My first instance is taken from Socrates’ dialectical—though fallacious—proof
that ‘to be punished justly is to be benefited’.

(10) 1. (αρα): When a man acts, there must always be a patient acted upon by the
agent.
2. (Spa) The patient does experience what the agent does: if, for instance, the
agent strikes, the patient must be stricken.
2a (καί) If the agent strikes hard or fast, the blow received by the patient
must be of like quality.
3. (αρα + question mark) The experience of the stricken is of the same qual­
ity as the action of the striker.
4. (ούκοΰν καί) If the agent bums, the patient must be burned.
4a (καί) If the agent bums severely, the patient must be burned in the same
way.
5. (ούκοΰν καί) If the agent cuts the patient must be cut,
5a (καί) If the agent cuts deep, the patient must be cut in the same way.
6. (συλλήβδην δή: introducing the conclusion of the inductive argument) The
quality of the patient’s experience corresponds to that of the agent’s action.
7. (πότερον ... ή) To pay a penalty (δίκην οιδόναι) is to suffer,
8. (ούκοΰν) at the hands of some agent: the inflicter of the punishment.
9. (δέ) He who rightly inflicts it punishes justly,
10. (0) acting justly.
11. (ούκοΰν) He who is punished suffers justly when he pays the just
penalty.
12. (δέ) Just things are fine and honorable.
13. (αρα+ period) The punisher does, the punished suffers what is honor­
able.
162 C.M .J. SICKING

14. (ούκοΰν) If the punished suffers what is honorable he suffers what is

good (the honorable being either pleasant or profitable).
15. (αρα + question mark) He who is punished justly suffers what is good.
16. (αρα+ question mark) He who is punished is benefited. (476b2-477a3)

Regarding this argument the following observations can be made:

• The fundamental statements upon which the argument will be based are introduced
by αρα and πότερον... ή, leaving the interlocutor free to answer as he sees fit:
(1) when there is an agent, there must always be a patient;
(2) the patient experiences what the agent does a n d (2a) the quality of the experience
will depend upon the quality of the action;
(7) to pay a penalty is to suffer—the last statement being necessary because the
results of the inductive argument as it is summarized in step 6 can only be applied to
the case of δίκην διδόναι if Polus accepts that δίκην διδόναι, like τύπτεσθαι and
κάεσθαι, is an instance of suffering something.
• Together with the three statements that have been agreed upon earlier on (steps 9 &
10 and 12)—all of them introduced by δέ—the statements (1), (2), (2a) and (7) pro­
vide all that is needed to enable Socrates to prove his point.
• We find αρα in those statements that confront Polus with the consequences of his
having agreed to an earlier statement: (3) the experience of the stricken is of the same
quality as the action of the striker, (13) the punisher does, the punished suffers what
is honorable; (15) he who is punished justly suffers what is good, and, finally,
(16) he who is punished is thereby benefited (see Section 4. below).
• Ούκοΰν is used by Socrates for switching from preliminary or subsidiary material
to its actual application. Steps (4) and (5) both feature ούκοΰν κ α ί , and are the
application of what has been inferred from the άρα-questìons underlying the argu­
ment as a whole to the instances of burning and cutting. By means of step (8)
Socrates returns to the argument as such, after having secured that δίκην διδόναι is
rightly regarded as an instance of suffering. In steps (11) and (14), ούκοΰν marks
the transition from preliminary δέ-steps which have been agreed upon earlier on
(together with what is inferred from them) to the argument at hand.
As far as I can see there is no essential difference between the contribution of
οΰν when it is found on its own and when it is incorporated in ούκοΰν: in both cases
οΰν serves the purpose of shifting the focus of attention, either from preliminary
material to a point this material owes its relevance to, or from explanatory material to
the main line of the argument.
The actual, so-called ‘logical’ function of both ούκοΰν and οΰν-utterances
varies according to the function of the statement in question for the argument as a
whole. As both Van Ophuijsen and I have attempted to show, it is of little or no use
to say, with Denniston (425ff.), that οΰν is «used as a connecting particle» express­
ing « p o s t h o c and (more frequently) p r o p t e r h o c , or anything between the two»—
from, that is to say, «proceeding to a new point, or a new stage in the march of
thought» to «inferential». Nor is it helpful—or even relevant—for our understanding
 PARTICLES IN QUESTIONS IN PLATO 163

of ούκοΰν to classify, with Denniston (434), ούκοΰν-questions as (a) «Strictly infer­

ential, ‘therefore’, ‘then’, (b) «Proceeding to introduction of minor or major premiss,
‘Now’» (c) «In enumeration of details, ‘Again’» (d) «Proceeding to a new point, or
a new step in the argument, ‘Now’, ‘Again’» and (e) «Introducing a disquisition for
which the interlocutor has declared himself ready or eager, ‘Well’.»7 This is at the
most an inventory of the widely differing roles that can be assigned to the c o n t e n t of
οΰν- and ούκοΰν-utterances in arguments, and/or of the characteristics of the sur­
rounding context. The basic value, however, that underlies the use of both οΰν and
ούκοΰν in argumentative contexts seems to be that they help the interlocutor in
determining how a statement (or question) relates to the steps that have been agreed
on in the preceding context.
My second instance of ούκοΰν is the only one in the G o r g ia s where it occurs in
a question posed not by Socrates himself but by one of his interlocutors (if, that is to
say, the substitution of ούκοΰν for ούκοΰν in (2) - (4) above is accepted):

(11) P o lu s : Then you think rhetoric is a routine?

S o c r a te s : Subject to your approval I do.
P o lu s : What routine?
S o c r a te s : One that produces gratification and pleasure.
P o l u s : ούκοΰν καλόν σοι δοκεΐ ή ρητορική είναι, χαρίζεσθαι οίόν τε
είναι άνθρώποις;
S o c r a t e s : Τί δέ, ώ Πώλε; (If I am to answer your question, I need some
further information): have you already learned from me what I consider
rhetoric to be, that you proceed to ask if I do not think it a fine thing? (462c4-
d2)

Here Polus is copying Socrates’ way of using ούκοΰν as a signal that this is what he
was leading up to—although he is, of course, once again jumping to a conclusion not
supported by Socrates’ reply, and therefore does not get the positive reply he hoped
for.
In (12), both Burnet and Dodds print interspaced ούκ οΰν instead of ούκοΰν,
apparently because Polus is repeating the formula that has been used by Socrates
(‘Then do you still deny that they do what they want?’). The point, however, seems
to be precisely that, in Polus’ view, he has already furnished the proof required from
him by Socrates (άλλα μ’ έλεγχε), because, to him, ä δοκεΐ αύτοΐς βέλτιστα είναι
is synonymous with ö βούλονται: ‘is not this «do what they want»?’:

(12) How can rhetoricians or tyrants possess great power in our cities,
S o c r a te s :
unless Polus proves to Socrates that they do what they want? (Πώς άν οΰν οί

7 Cf. Kiihner-Gerth 2. 163-4: «ούκοΰν wird zuerst und ursprünglich in der Frage gebraucht und
bedeutet nonne igitur, nonne ergo. In einer solchen Frage liegt das ganze Gewicht des Gedankens
auf dem syllogistischen οΰν; aus der Folge selbst ergibt sich, dass der fragweise ausgesprochene
Gedanke zu bejahen sei; die Negation ist bloss deshalb hinzugefügt, damit die bejahende Antwort
des anderen unzweifelhaft gesetzt und als von selbst folgend bezeichnet werde.»
164 C .M .J. SICKING

ρήτορες μέγα δύναιντο ή οι τύραννοι έν ταΐς πόλεσιν, έά ν μή

Σωκράτης έξελεγχθή ύπό Πώλου δ τι ποιοΰσιν ά βούλονταν, )
Polus: This fellow ,..
Socrates: οϋ φημι ποιεΐν αυτούς ö βούλονται' αλλά μ' έλεγχε.
Polus: Did you not just now admit that they do what seems best to them?
Socrates: Yes, and I still admit it.
Polus: ούκ ούν ποιοΰσιν ä βούλονταν,
Socrates: οϋ φημι.
Polus: When they do what seems good to them?
Socrates: Yes.
Polus: What you are saying is shocking and fantastic, Socrates. (478a8-bl0)

Regarding punctuation it will by now be clear that, while disagreeing with Dennis-
ton’s criterium for distinguishing ούκοΰν-questions from those with ούκουν, I
would agree with him in printing a question mark after all ούκοΰν-utterances of the
type I have just exemplified, ούκοΰν-questions—like άλλο τι οϋν-questions—differ
from οϋκουν-questions in that the latter are rhetorical questions expressing the
questioner’s disbelief that his interlocutor really means what he says, whereas the
former typically are assent-seeking or apodeictic questions, and as such are to be
distinghuished from veritable yes/no questions introduced by αρα and Sp' ούν.

3. ά ρ α a n d <?.p' ο ύ ν8

In

(13) Socrates: Or rather, Gorgias, tell us yourself in what art you are expert and
what we should call you.
Gorgias: The art of rhetoric, Socrates.
Socrates: (άρα) Then we must call you a rhetorician?
Gorgias: Yes, and a good one, Socrates, if you really want to call me what I
boast myself to be.
Socrates: That is indeed what I want.
Gorgias: Then call me so.
Socrates: (ούκοϋν) Are we to say that you can make rhetoricians of others
also?
Gorgias: That is the profession I make both here and elsewhere.
Socrates: TAp' ούν έθελήσαις άν, co Γοργία, ώσπερ νυν διαλεγόμεθα,
διατελέσαι τό μεν έρωτών, το δ' άποκρινόμενος, το δέ μήκος των λόγων
τούτο, οΐον καί Πώλος ήρξατο, εις αϋθις άποθέσθαι; (449a2-b8),

8 δρα: 47b9, 4 49el, 453dl, 453c7, 454d5, 459c4, 462b6, 467e7, 476a8, 476b4, 476b5, 477a5,
499e7,506el.
δρ' ούν: 449b4,449e6,450c3,463dl, 4666a9,467el, 468d5, 477a8,477d2,478a6,478b7,478c3,
479c8,479d7,488e7,49Se6,498c6,499dl, 499d4,500a4,510e4,513e5.
 PARTICLES IN QUESTIONS IN PLATO 165

from a viewpoint of content, there is no ‘logical’ relationship between Socrates’ Sp'

οΰν-question and the preceding questions. The presence of οΰν is due to Socrates’
referring back to an exchange he and Polus have had earlier on, when Polus proved
unable to restrain himself from embarking on a lengthy encomium of the art of
rhetoric, instead of answering the question he was asked; Socrates’ reaction was that
he would prefer to question Gorgias himself (448d7-8). The effect of οΰν actually is
that it, so to speak, reduces Socrates’ first short exchange with Gorgias to a kind of
preliminary test he wanted him to pass before taking his chance with him as a con­
versation partner.9 Now that Gorgias has more or less proved himself capable of
meeting Socratic standards, Socrates wants to know if he is prepared to continue (and
to finish the conversation) along the same lines. It is, of course, a matter of courtesy
that Socrates—as we may gather both from the optative and from his using Spa—
leaves Gorgias free to answer as he likes.
Having secured Gorgias’ willingness to give concise answers, Socrates proceeds
to seriously questioning him about his expertise. He asks him what is the field of
rhetoric, and shows him that his first answer leaves much to be desired:

(14) Socrates: (περί τί) What is the field of rhetoric?

Gorgias: Words.
Socrates: (ποιους) Words of what kind? αρα o'i δηλοΰσι τούς κάμνοντας,
ώς αν διαιτώμενοι ύγιαίνοιεν;
Gorgias·. No.
Socrates·, (ούκ αρα + period) Then rhetoric is not concerned with every kind
of words.
Gorgias: Certainly not.
Socrates: (άλλα μην;) Yet it makes men able to speak?
Gorgias: Yes.
Socrates: (ούκοΰν;) Able to think also about the matter of their discourse?
Gorgias: Of course.
Socrates: (αρ’ οΰν;) Does the science of medicine, which we have just men­
tioned, make men able both to think and to speak about their patients?
Gorgias: Assuredly.
Socrates: (άρα+ period) Then medicine also, it seems, is concerned with
words.
Gorgias: Yes.
Socrates: (0) Words about diseases?
Gorgias: Certainly. (449el-450a5)

Once again, Spa indicates that Socrates is open to any answer Gorgias will produce.
The suggestion that the ‘words’ are words about sickness and health is as good as

9 For the reader, the question echoes the introduction to the dialogue (447b9-cl), where Socrates
wondered whether Gorgias would be willing to converse with him: άλλ' αρα έθελήσειεν αν ήμιν
διαλεχθηναι;
166 C .M J . SICKING

any other—although it is, perhaps, not without significance that Socrates proposes
precisely this differentia·, the analogy between philosophy and medicine, and between
physical and moral health, will turn out to be one of the central themes o f the dia­
logue (cf. also 448b4). As δρα indicates, Socrates does not expect an affirmative
answer, but it is far from impossible that he would have accepted, or even welcomed,
it as a basis for further conversation.
Socrates then makes Gorgias admit that maintaining that rhetoric is a τέχνη
‘about words’ implies that its practitioners will be capable not only of speaking but
also of thinking rationally about the matter of their discourse. With the Sp’ οΰν-
question he returns to the analogy with medicine in its enriched version: is Gorgias
prepared to endorse the premiss that medicine will make men able both to speak and
to think about their patients? On the basis of this, the final conclusion that just like
rhetoric, medicine also is περί λόγους, besides making clear that Gorgias ’ definition
is too wide, also implies that, just like any other art, rhetoric will have to make its
practitioners capable of understanding the ‘principles’ of its subject matter. This
implication will turn out to be of fundamental importance for the main issue of the
dialogue.
One final example of Spa and αρ' οΰν can be taken from the continuation of a
passage I have quoted above [(10)]. There, after having Polus made admit that «he
who is punished is benefited», Socrates proceeds to find out what kind of benefit
Polus is thinking of:

(15) Socrates: νΑρα ήνπερ έγώ υπολαμβάνω την ώφελίαν; βελτίων την ψυχήν
γίγνεται, είπερ δικαίως κολάζεται;
Polus: Είκόςγε.
Socrates: (άρα;) Then he who is punished is rid of evil in the soul?
Polus: Yes.
Socrates: TAp' ουν του μεγίστου άπαλλάττεται κακού; Look at it in this
way (ώδε δέ σκόπεν): In the fabric of man’s material estate do you see any
other evil than poverty?
Polus: No, only poverty.
Socrates: Ti 5’ έν σώματος κατασκευή; Would you say its evil is weakness
and sickness and ugliness and such things?
Polus: I would. (477a5-b8)

Here also, both the Spa- and the Sp' οΰν-question can be seen to be intended as veri­
table yes/no questions. The kind of benefit Socrates is referring to needs further
iV
specification: is his soul made better if he is rightly punished? Polus will agree to
this, this will provide Socrates with a premiss for the further statement that «(άρα) he
wo is punished is rid of evil in the soul» (cf. 475c7-9). This in turn will provide the
basis for asking Polus the question Socrates has been aiming at for a long time:
would Polus agree or not that «evil in the soul» is «the greatest of evils»? For the
 PARTICLES IN QUESTIONS IN PLATO 167

time being the question is left open: Polus is not supposed to answer it before it has
been subjected to a separate investigation, introduced by ώδε δέ σκόπει.
Only at 477e4-6 he is invited to answer it on the basis of the outcome of the
investigation:

(15a) Ή αδικία άρα κ α ί ή ακολασία καν ή άλλη ψυχής πονηρία μέγιστον

των οντων κακόν έστιν;
- Φαίνεται.

4. αρα

The point of departure of my discussion of αρα in questions is the treatment of αρα

by Van Ophuijsen, who convincingly defines (102) its contribution to step by step
arguments as conveying «that the present statement has to be accepted on the
strength of the preceding ones as soon as these are granted». We may render its
contribution in English by expressions such as: ‘if this is so’, ‘it appears’, ‘if this is
granted’. This may be illustrated from two passages I have quoted earlier on.
In (10) steps 3 (‘if this is so, the quality of the blow received by the stricken
depends on the quality of the blow given by the striker’), 13 (‘if it is granted that just
things are honorable, then the man who is punished [justly] suffers what is honor­
able’), 15 and 16 are cases in point. In [14] we have similar instances both of οΰκ
αρα and of αρα.
I am, then, fully prepared to follow Van Ophuijsen (101) in taking it that, in
Plato, αρα is far from being «used practically as a variant for οΰν and δή»
(Denniston 40), but, on the contrary, «is more apt than both οΰν and δή to be used in
cases of strictly formal connexion»: «αρα ties the acceptance of a proposition to an
assumption in the context» (Van Ophuijsen 87). Thus, άρα is, so to speak, on the
lower side of a gliding scale that goes from veritable yes/no questions via various
other question types (among them ούκοΰν-questions, questions with simple οΰν,
with δέ. S e e .) to those questions where the interlocutor is given to understand that he
is supposed to have committed himself to confirming the suggestion offered to him.
άρα-questions submit «for confirmation or denial a suggestion based on the preced­
ing statement or formula of assent» (Van Ophuijsen 116). What makes them differ­
ent from questions that are introduced by other particles is precisely the nature of the
relationship between the content of the άρα-utterance and the statement(s) confirmed
earlier on by the interlocutor. By using άρα the questioner suggests that his conver­
sation partner must either accept the predication based on the preceding ώμολογη-
μένα or invalidate the logic behind the questioner’s suggestion so as to force him to
vindicate his argumentation or to abandon this line of reasoning.
The one problematic aspect of άρα in the present context is the remarkable fact
that, in our editions, some άρα-utterances are printed with a full stop, whereas after
others a question mark is inserted. I have spent some time on trying to find out why
this is so, but have not succeeded in tracing any consistent criteria that may have
168 C.M.J. SICKING

made Burnet—who is without exception followed by Dodds—decide in favour of the

one or the other. In 476b3ff. [= (10)], I see no relevant difference between steps 3,15
and 16 on the one hand and step 13 on the other, that would make one print a period
in the former, and a question mark in the latter, nor can I discover any characteristic
of the two instances of (ούκ) αρα in 449elff. [= (14)] that would justify printing a
full stop rather than a question mark. If it is correct that the force of dpa is to be
defined in terms of the ‘acceptance’ rather than of the ‘truth’ of the statements
involved (cf. Van Ophuijsen 139), printing a period after (a selection of) αρα-ques-
tions would seem to betray a lack of understanding of Socratic dialectic: the mere fact
that Socrates is offering a proposition that is based on statements previously agreed
on does not imply that the matter is closed. Just as in the case of other question-
statements, the content of άρα-questions has to rise to the status of ώμολογημένον
before it can be used as a basis for further argument. It is, perhaps, a matter of taste
whether one wants to make this visible by printing a question mark in the text; if,
however, an editor would prefer— as I would not—printing a period he might at least
be expected to do so consistently.

5. τ ί οΰν, τί δέ, τ ί δη, τί οΰν δή and τί ποτέ

In the final part of this paper I will offer some observations on τί οΰν;101τί δέ;11 τί
δή;12 τί οΰν δή;13 τί ποτέ;14
According to Kühner-Gerth, questions introduced by τί(ς) οΰν are used in dia­
logue either «wenn der Fragende in Beziehung auf eine gethane Äusserung eines
anderen mit Lebhaftigkeit (my italics) Aufklärung verlangt» or «folgernd», in a way
that is to be compared with Latin quis igitur . . . ? (K-G 2,161-2). I would, however,
submit that there is nothing in τ ί οΰν that makes it especially apt for being used in
lively questions or for inviting one to affirm a conclusion drawn from some preced­
ing statement.
In the Gorgias there are 11 instances of τί οΰν, either standing on its own or
introducing a complete sentence. In some cases it is used by one of Socrates’ con­
versation partners, as in:

(16) Socrates: If you wish to question me, ask me what part of flattery I claim
rhetoric to be.
Polus: I will. Answer me: what part?
Socrates: I wonder whether you will understand my answer. Rhetoric in my
opinion is the semblance of a part of politics.
Polus: Τί οΰν; καλόν ή αισχρόν λέγεις αύτην είναι;

10452a7,452c4,460bl, 463d3,466a4,478b3,498a7,504b8,509c9, 516a4,516bl0.

11 34 occurrences in the Gorgias.
12 448el. 469a3.470a4.486e4.
 PARTICLES IN QUESTIONS IN PLATO 169

Socrates: Bad—for evil things I call bad—if I must answer you as though
you already understand what I mean. (463c6-d5)

Here, Polus’ problem with Socrates’ answer to his first question, apparently, is that
he cannot understand its relevance for what is, for him, the subject-matter at hand:
how is Socrates’ answer going to contribute to answering the question whether
rhetoric is to be called good or bad? So, if one wishes to paraphrase the value of τί
οΰν, it will have to be by something like: ‘What is the relevance of this for the issue
at hand? What I want to know is: do you mean that rhetoric is good, or bad?’
The reverse of this can be seen in (17), a passage that will provide us at the same
time with some characteristic instances of τί δέ:

(17) Socrates: The good is not the same as the pleasant, nor the evil as the painful.
For we cease from the one pair at the same time, but not from the other,
because they are distinct. How then (πώς συν marks the return from the
explanation to the main line of the argument) could the pleasant be the same
as the good, or the painful as the evil? (...) Do you not call good people by
that name because of the presence in them of things good, just as you call
beautiful those in whom beauty is present?
Callicles: I do.
Socrates: Τί δέ; Do you call fools or cowards good men? You did not just
now, but it was the brave and the wise, or do you not call these good?
Callicles: I do.
Socrates: Τί δέ; Have you ever seen a silly child enjoying pleasure?
Callicles: Yes.
Socrates: And never seen a silly man enjoying pleasure?
Callicles: Yes, I suppose so. αλλά τί τούτο,
Socrates: Nothing, just answer.
Callicles: Yes.
Socrates: Τί δέ; And a sensible man experiencing pain or pleasure?
Callicles: Yes.
Socrates : Which (πάτερου δέ) feels more pain or pleasure, the sensible or the
fool?
Callicles: I do not think there is much difference.
Socrates: That will do. Have you ever seen a coward (δέ) in battle?
Callicles: Of course.
Socrates: Ti ouv; When the enemy retreated, which of the two seemed more
to rejoice, the cowards or the brave?
Callicles: Both equally, I think, or if not, pretty much so.
Socrates: It doesn’t make any difference. We may leave it at that: what I want
to know is whether cowards too feel pleasure, just like the brave (Ούδέν
170 C.MJ. SICKING

διαφέρει, χαίρουσιν δ' οΰν15 καί οί δειλοί; - σφόδρα γε, - Κ αί οί

άφρονες, ώς έοικεν). (497d6-498b4)

What makes this exchange remarkable is that Socrates is posing a whole series of
seemingly unconnected questions, all of them introduced by τί δέ; or containing just
δέ. Only at 498b7 ούκοΰν will mark the question which at last reveals the application
of the scries: ‘Then fools and wise, cowards and brave feel pain and pleasure to a like
degree, as you say, but the coward more so than the brave?’ (Ούκοΰν λυπούνται
μέν καί χαίρουσιν καί οί άφρονες καί οί φρόνιμοι καί οί δειλοί κ α ί οί
ανδρείοι παραπλήσιος, ώς συ φής, μάλλον δέ οί δειλοί των ανδρείων;)
When asking Callicles whether he has ever seen a coward in battle, Socrates is
pushing to extremes the principle of making each and every small step into a separate
question. By using, for once, τί οΰν instead of τί δέ he makes clear that he is still
observing what, in pragmatics, is called the «Maxim of Relevance»: ‘Why am I ask­
ing this? I want to know which of the two seemed more to rejoice, when the enemy
retreated, the cowards or the brave.’
The difference between τί δέ and τί οΰν as they are used in this passage seems
to be that the former opens a new unit of development by introducing a new question
about a new topic, whereas the latter signals that the preceding question derives its
relevance from its being a stepping stone to the one that is to follow.
These observations must suffice to show why I would not subscribe to Dennis-
ton’s (175) description of the uses of so-called ‘elliptical’ τί δέ: «(a) (The common­
est.) Expressing surprise or incredulity, and usually introducing a further question
(‘What?’); (b) With the connective force of δέ more prominent: ‘And what of that?’
‘Well’: O f course’: ‘Que voulez-vous?’ A shrug of the shoulders, (c) As a formula
of transition: ‘And what (of this that follows)?’» Nor can I follow Kiihner-Gerth (2,
263), who distinguish two ways of using τί δέ: «(...) häufig im Dialoge, das allein­
stehende τί δέ; in leidenschaftlicher Frage = quid v e r ö l doch oft auch bei einem
Übergange und einer Fortsetzung der Rede = quid a u t e m l »16
Denniston’s (176) one instance of τί δέ; from the G o r g i a s is

(18) S o c r a te s : (...)if you can mention any such orator, why do you not tell me his
name at once?
C a llic le s : 1 cannot give you the names of any such orators living today.
S o c r a t e s : Τί δέ; Can you name any of the orators of old, from the time of
whose first appearance to address the public, the citizens (...) are said to have
been improved?

1:5 Δ' οΰν marks the transition both to a new topic (δ') and to what is relevant now (οΰν).
16 Cf. also 2,262-3: «Ebenso bei lebhaften oder leidenschaftlichen Fragen oder bei Ausrufungen
des Unwillens, wo δέ einen Gegensatz zu einem aus dem Vorhergehenden oder Folgenden oder aus
dem ganzen Zusammenhang zu ergänzenden Gedanken bildet (Anm.: «Allerdings liegt diese
Beziehung zuweilen sehr versteckt im Gedankenzusammenhange; aber an den meisten Stellen
springt sie in die Augen») Z, 123 τις δέ σΰ έσσι ...; wer bist denn aber du? wo a u s dem
Folgenden (my italics) der Gedanke zu entnehmen ist: du wagst mir entgegenzntrelen.»
 PARTICLES IN QUESTIONS IN PLATO 171

Ti δέ; Have you never been told that Themistocles was a good man,
C o llid e s :
and Cimon and Miltiades, andPericles (...)? (503bl-c3)

This is very much like the above instances in that, by means of τί δέ, Socrates
switches from one topic, contemporary politicians, to a new one, the politicians from
the past, whereafter Callicles comes up with a suggestion of his own. If thus anal­
ysed, both instances represent one and the same type of conversational move,
whereas, if we were to follow Denniston, they would have to be assigned to two dif­
ferent categories.
So it seems to me that there is a sound basis for leaving such intangibles as
‘surprised’, ‘incredulous’, ‘indignant’ or ‘passionate’ for what they are, and distin­
guishing between τί οΰν and τί δέ by observing that they characterize different rela­
tions between the question that is to follow and the preceding context.
Τί δέ is mainly used in two characteristic situations:
• In those cases where τί δέ precedes a question that is a member of a series it indi­
cates that the speaker is about to ask a new question that will be about a new topic
(‘Yet another question: ...’, ‘Here my next question: ...’).
• When τ ί δέ precedes a question that represents a reaction to something asked or
said by someone else it invites the conversation partner, so to speak, to ‘open a new
paragraph’, so as to supply further explanation or information, A case in point is

(19) P o l u s : If you wish, Chaerephon, make the experiment upon me, for Gorgias, I
think, is played out: he has already spoken at great length.
C h a e r e p h o n : Τί δέ, ώ Πώλε; do you think you could answer better than
Gorgias? (448a6-10; 447b4 is a similar case.)

In instances such as this last one there may seem to be some ground for supposing
that the questioner is inspired by surprise or incredulity. This is, however, not a con­
sequence of τί δέ having been used as such: it can at most be seen as a side-effect in
so far as signalling that, in order to understand what has been said, one is in need of
further information, will, after all, often be inspired by disbelief or surprise.
Regarding τί οΰν, a similar dichotomy can be made:
• In instances such as (17), quoted above, τί οΰν indicates that Socrates’ next ques­
tion will explain what his preceding question was aiming at: ‘I asked this with a view
to my next question: ... ’
• In other instances, τί οΰν serves the purpose of expressing that the questioner is
puzzled about the relevance of something that just has been said. In this case, we also
find τί οΰν δή, as, for instance, in20

(20) Then, listen, Gorgias. I am convinced that if there is any man who
S o c r a te s '.
in a discussion with another is anxious to know what is the real subject under
discussion, I am such a man, and I am confident that you are.
172 C.M.J. SICKING

Gorgias: τί ούν δή, <δ Σώκρατες; (453b4).17

You will surely understand that I do not see what you are aiming at,
Socrates.

As to the contribution of δή: it conveys Gorgias’ assumption that Socrates will be

aware of the reason for his asking for further explanation. We may compare

(21) Socrates: (,..) του δή ένεκα λέγω ταΰτα; cm νΰν έμοί δοκέΐς συ ού πάνυ
ακόλουθα λέγειν ουδέ σύμφωνα οΐς το πρώτον έλεγες π ερ ί της
ρητορικής- (457el-3),

where του δη ένεκα occurs in the course of a monologue by Socrates, who is, so to
speak, anticipating the possibility that Gorgias will have a problem with the relevance
of what he is saying.18 In

(22) Socrates: Do you mean justly or injustly?

Polus: Whichever way, is he not to be envied in any case?
Socrates: Hush! (εύφήμει, ώ Πώλε)
Polus: τί δή;
Socrates: Because we should not envy the unenviable and miserable, but pity
them. (469al-5),

τί δή conveys Polus’ expectation that Socrates will understand his inability to imag­
ine what was wrong with what he has said: could he, please, explain himself?
Δή,1920then, is not only much less frequently used in questions than οΰν, αρα
and δε: it also differs from these in that its purpose is not to establish the position of
the questions involved in the wider context of an argument, τί δή belongs to the area
of «interpersonal pragmatics», and is best read as signalling the involvement of the
questioner in maintaining a basis for communication, a signal that his being puzzled
is not to be taken as indicating a lack of cooperativeness.
In this respect, τί δή belongs to the same area as τί ποτέ.20 By using τ ί ποτέ a
conversation partner may indicate that he himself finds it difficult to think o f a satis­
factory answer to his question. Thus in

(23) Τί ούν δή ποτέ τάς αλλας τέχνας ού ρητορικός καλεΐς, ούσας περί
λόγους, είπερ ταύτην ρητορικήν καλεΐς, ή αν ή περί λόγους; (450b3-5)

by means of ποτέ, Socrates conveys that, for Gorgias, the only way out is to concede
that his definition of the field of rhetoric is unsatisfactory. An attempt at bringing out

17 Cf. 450b3,452d9,505c7,515el.
18 Denniston 239 «After saying something it is natural to justify il» seems to me completely irrel­
evant
19 Cf. 448el, 453cl, 469a3,470a4, 480b6,486cl, 486e4,502bl.
20 Cf. 455a8,456a6,488cl, 489d7,490al, 502c2.
 PARTICLES IN QUESTIONS IN PLATO 173

as explicitly as possible the respective contributions of οΰν, δή and ποτέ yields

something like: ‘when I apply what you have said so far to the problem at hand
(οΰν), you will find it only natural (δή) that I ask a question which you very well may
find difficult to answer (ποτέ): why ...’ To offer a less cumbrous rendering: ‘The
question now to be asked, Gorgias, is as natural as it is difficult: as these arts have to
do with words, why not call them by the name of ‘rhetoric’—if, that is to say,
(ε'ίπερ) you call by that name any art that is concerned with words. ’
This way of accounting for τί οΰν δή ποτέ seems preferable to choosing
between the devil of locating both οΰν and δή somewhere on the scale «between p o s t
h o c and p r o p t e r h o c », and the deep blue sea of taking it that Socrates’ surprise or
even indignation at Gorgias’ inconsistency is such that he gives vent to it in at least
three different ways 21

6. Conclusion

A description of Plato’s use of particles for introducing questions must distinghuish

between at least three categories:
♦Particles, such as Spa and πότερον ... ή, that introduce veritable questions in the
sense of questions that leave the conversation partner free to answer as he sees fit: he
may say ‘yes’ or ‘no’, propose a modification of the question before answering it, or
even refuse to answer at all. In many cases the questions involved concern predica­
tions that will serve as a point of departure22 of a longer argument: the conversation
partner may answer as he likes, but he will have to stick to the position he has chosen
and accept the consequences of his choice.
• Particles such as ούκοΰν, οΰν, dpa and δε serve the purpose of articulating the
argument by showing how a question relates to what has preceded. We will not
achieve much when we seek to define the respective functions of ούκοΰν, οΰν, dpa
and δέ by a logic-inspired classification of the questions involved, and observing that,
for instance, οΰν (like dpa) is used in deduction, in the conclusion of a syllogism, in
induction and in the statement of a law (Des Places p a s s i m ) . 23 Nor is it helpful to
probe into the inner feelings of the questioner, asking ourselves whether he is sur­
prised, angry, impatient or anything of the sort. It seems more to the point to note that
ούκοΰν, οΰν and δέ regard only the purposes the speaker has with the question
involved in its context, whereas dpa, moreover, «ties the acceptance of the proposi­
tion to an assumption in the preceding context». Regarding the difference between
(ούκ)οΰν and δέ, I have submitted that the respective contributions of οΰν and δέ to
questions are m u t a t i s m u t a n d i s similar to their contribution to a piece of connected
speech. It is to be noted that δέ is also used for the purpose of incorporating state­
ments agreed upon in the preceding conversation into the argument at hand.

21 Cf. Kühner-Gcrth 2, 161: «Auch τ ι δή; allein = quid tandem? mit Verwunderung: wie in aller
Welt? = das ist offenbar nichL möglich.»
22 Since 1 am not occupied with an analysis of logical procedures, technical terms, such as
‘premiss’, would be out of place.
23 Cf. Van Ophuijsen 87,89.
174 C.M.J. SICKING

• So-called ‘interactive’ particles such as δή and ποτέ that are used as instruments by
which a questioner may direct (or manipulate) the interaction between himself and his
interlocutor.24

Select Bibliography

Bakker, E.J.
1993 ‘Boundaries, Topic and the Structure of Discourse: An Investigation of the
Ancient Greek Particle d f . S t u d i e s i n L a n g u a g e 17, 275-311
Denniston, J.D.
19542 T h e G r e e k P a r tic le s . Oxford
Devine, A.M. & L.D. Stephens
1994 T h e P r o s o d y o f G r e e k S p e e c h . New York & Oxford
Kroon, C.
1995 D i s c o u r s e P a r t i c l e s i n L a tin . A S t u d y o f nam, enim, autem, vero a n d at.
Amsterdam
Kühner, R. & B. Gerth
1898-1904 A u s fü h r lic h e G r a m m a tik d e r g r ie c h is c h e n S p r a c h e . Zweiter Teil,
Satzlehre, Zweiter Band. Hannover & Leipzig
Leech, G.N.
19853 P r in c ip le s o f P r a g m a tic s . London & New York
Mastronarde, D.
1979 C o n ta c t a n d D i s c o n t i n u i t y . S o m e C o n v e n tio n s o f S p e e c h a n d A c t i o n o n th e
G r e e k T r a g ic S ta g e . Berkeley & Los Angeles & London
Places, E. des
1929 E tu d e s s u r q u e lq u e s p a r tic u le s d e lia is o n c h e z P la to n . Paris
Sicking, C.M.J. & J.M. van Ophuijsen,
1993 T w o S t u d i e s i n A t t i c P a r t i c l e U s a g e . L y s i a s & P l a t o . Leiden. Part I:
‘Devices for Text Articulation in Lysias I and ΧΙΓ, by C.M.J. Sicking; Part
Π: ΌΥΝ, APA, ΔΗ, TOINTN: The Linguistic Articulation of Arguments in
Plato’s P h a e d o by J.M. van Ophuijsen

24 Cf. Sicking 51.

 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES’

IRENE J.F . DE JONG

Universiteit van Amsterdam

I n tr o d u c tio n

My point of departure is the following passage in Herodotus:

(1) ... τό μέν 'Αττικόν κατεχόμενόν τε και διεσπασμένον έπυνθάνετο ό

Κροΐσος ύπό Πεισιστράτου του Ίπποκράτεος τούτον τόν χρόνον τυραν-
ν ε ύ ο ν τ ο ς Ά & η ν α ί ω ν ^ Ί π π ο κ ρ ά τ ε ϊ γάρ έ ό ν τ ι ιδιώτη κ α ϊ θ ε ο ψ έ ο ν τ ι τ ά
'Ολύμπια τέρας έγένετο μέγα· (1.59.1)),
‘... Croesus learned that the Attic people was held in subjection and divided
into factions by Pisistratus, the son of Hippocrates, who at that time was
tyrant in Athens. To this Hippocrates a great marvel befell when, as a private
person, he was in Olympia to see the games.’

At first sight γάρ hardly seems to be fulfilling its usual function, i.e., to introduce an
explanation of what precedes. Denniston (1954: 60) includes the passage in subcate­
gory 5 of the explanatory use of γάρ (Other examples do not fall under any of the
above headings’); this is obviously a repository for difficult cases, which does not
help us very much. (Note also that the translator of the Loeb edition, A.D. Godley,
whom I quote in (1), does not translate the particle at all.) However, the correct inter­
pretation of γάρ can be found by simply reading on. It turns out that the particle
occurs at the opening of an embedded narrative, extending over five chapters, which
tells us how Pisistratus became tyrant of Athens: his father Hippocrates was warned
in Olympia by Chilon not to have children; nevertheless he fathered a son, Pisistratus,
who in two moves managed to establish himself as tyrant of Athens. In chapter 65
the Herodotean narrator picks up his main story with a characteristic e p a n a l e p s i s :
τούς μέν νυν ’Αθηναίους τοιαύτα τόν χρόνον τούτον έπυνθάνετο ό Κροΐσος
κατέχοντα,... (‘Regarding the Athenians Croesus heard that these kinds of things
were occupying them at that time, ...’). In other words, the narrative a s a w h o l e
constitutes an explanation of ύπό Πεισιστράτου ... ’Αθηναίων in 59. Γάρ fulfils its
usual, explanatory, function, but the particle has a larger scope. (It is typical of

* I wish to thank A. Rijksbaroit and C.M.J. Sicking for their comments, and Mrs. B.A. Fasting for
her correction of my English.
 IRENE J.F. DE JONG

Denniston that he interprets particles largely at the level of the sentence rather than of
the text.)
The object of this paper is to argue that γάρ is in fact often used at the opening
of embedded narratives and that this use is widespread enough to have warranted a
separate subcategory in Denniston. (It is usually the other way around, and Dennis-
ton is critized for his abundance of subcategories.) I will also try to explain how γάρ
came to be used in this way. The key to this understanding lies in Homer, and it is
with this author that I will start my discussion.

Homer

In his recent study of γάρ Sicking argues, in my view convincingly, that Denniston’s
distinction between an explanatory and a causal use of γάρ is unfounded: ‘the pur­
pose of sentences introduced by γάρ is primarily explanatory’ (1993: 23). The
explanation can take a variety of forms: why X happens, why somebody does Y, why
the speaker says Z, etc. What interests me here is the temporal relationship between
the explanation hosted by γάρ and the statement which it explains. It turns out that
we find all three possibilities: the content of the γάρ-clause can be simultaneous,
posterior, or anterior. Let me give an example of each possibility, starting with simul­
taneity:234

(2) αύτάρ έπεί κατά τέκνα φάγε στρουθοιο καί αυτήν,

τον μεν άρίζηλον θηκεν θεός δς περ έψηνε'
λάανγάρ μιν έθηκε Κρόνου πάϊς άγκυλαμήτεω (//. 2.317-9)
‘But after he [the snake] had eaten the sparrow herself and her children, the
god who had sent him made him an evident sign. For the son of devious
devising Cronus turned him into stone.’

Posteriority:

(3) “θωρήξαί έ κέλευε κάρη κομόωντας 'Αχαιούς

πανσυδίη· νυν γάρ κεν έλοι πόλιν εύρυάγυιαν
Τρώων” (//. 2.11-13)
‘“Urge him to arm the long-haired Greeks in full force, for now he could
take the city of Troy’”

Anteriority:

(4) την μεν Άχιλλήος ρηξήνορος υίέϊ πέμπεν

έν Τροίη γάρ πρώτον ύπέσχετο καί κατένευσε
δωσέμεναι ( O d . 4.5-7)
‘He [Menelaus] was sending away his daughter in marriage: for already in
Troy he had promised and consented to give her’
 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES 177

This last use of γάρ, to introduce information relating to the past, is fairly widespread
in Homer; roughly 20% of the instances are anterior (as against 75% simultaneous
and 5% posterior). Often the anteriority is made explicit by the use of temporal mod­
ifiers like ήδη (e.g., I I . 3.205), ποτέ (e.g., I I . 6.216), πάρος (e.g., II . 11.111), πρίν
(e.g., O d . 11.484), ού ... πω (e.g., II . 1.154), or χθιζός (e.g., II . 1.423).
This ‘anterior’ γάρ is often employed in the narrative device of the epic regres­
sion: the narrator mentions an event, then goes back in time step by step until a cer­
tain point, and then narrates the events again, often in more detail, in their normal pro­
gressive order (schematically: D-C-B-A-A’-B’-C’-D’).1A famous example is the
beginning of the Ilia d :

(5) D Τις τ' op σφωε θεών έριδι ξυνέηκε μάχεσθαι;

Λητούς καί Διός υιός' (8-9)
‘What god was it then set them together in bitter collision?
Zeus’ son and Leto’s, Apollo,’2
C ό γάρ βασιλήϊ χολωθείς
νοΰσον άνά στρατόν ώρσε κακήν, όλέκοντο δέ λαοί, (9-10)
‘who in anger at the king drove
the foul pestilence along the host, and the people perished,’
B οϋνεκα τον Χρύσην ήτίμασεν άρητηρα
Άτρεΐδης (11-12)
‘since Atreus’ son had dishonoured Chryses, priest of Apollo,’
A ο γάρ ήλθε θοάς έπί νηας 'Αχαιών I ... I ... I
καί λίσσετο πάντας 'Αχαιούς,
’Ατρείδα δέ μάλιστα δύω, κοσμήτορε λαών (12-16)
‘when he came beside the fast ships of the Achaeans ...
... and supplicated all the Achaeans,
but above all Atreus’ two sons, the marshals of the people:’
A’ Chryses’ supplication (17-21)
B’ Agamemnon treats Chryses dishonourably (22-33)
C’ Apollo sends the plague and men start to die (34-52)
D’ Achilles and Agamemnon quarrel (53-305)

The narrator spirals back in time with the help of γάρ-clauses and then recounts the
same events again, now in more detail (288 lines against 8).3 Γάρ in 9 and 12 fulfils
its normal explanatory function: the narrator explains why Apollo set up a quarrel
between Achilles and Agamemnon and how Chryses came into contact with
Agamemnon. In passing, I would like to draw attention to οΰνεκα in 11; why do we
not find γάρ here? In my view, the narrator has chosen to present a causal connection1*3

1 See Schadewaldt (1938: 84) and Slater (1983). Epic regressions may also proceed by relative
Jronouns, e.g., Od. 4.125.
Translation of R. Lattimore, Toronto 1951.
3 Other examples of γάρ in epic regressions: II. 11.688, 690; 24.602 (καί γάρ); Od. 8.79; and
19.275,276.
 IRENB J.F . DE JONG

(‘why was Apollo angry?’) not in the form of an explanation, but in the form of
Apollo’s own motive: the god was angry, because in his eyes Agamemnon, rejecting
Chryses’ plea, had dishonoured the priest. In other—narratological—words, the
causal link is provided here in the form of Apollo’s embedded localization. This
analysis is confirmed by the presence in the οϋνεκα-clause of character language:
the verb άτιμάω occurs 15 times in a speech, only twice in narrator-text (here and in
O d . 21.99, where we are dealing with a clear narratorial intervention).4 While γάρ (in
Homer) gives us an explanation by the narrator, ουνεκα gives us a character’s
motive.5
A variant of this regressive form of storytelling is exemplified by the following
passage, in which the narrator mentions an event and then i n o n e s t e p goes back in
time to recount in detail how this event came about (schematically: E-A-B-C-D-E’):

(6) E 'Άδρηστον δ’ άρ' έπειτα βοήν άγαθός Μενέλαος

ζωόν έλ'·
A ϊππω γάρ οί άτυζομένω πεδίοιο
όζω έν ι βλαφθέντε μυρικίνφ,
Β άγκύλον άρμα
άξαντ’ έν πρώτω ρυμω
C αύτώ μέν έβήτην
προς πόλιν, ή περ οί άλλοι άτυζόμενοι φοβέοντο,
D αυτός δ' έκ δίφροιο παρά τροχόν έξεκυλίσθη
πρηνής έν κονίησιν έπ ί στόμα·
Ε’ πάρ δέ οί έστη
Άτρεΐδης Μενέλαος έχων δολιχόσκιον έγχος. (II. 6. 37-43)6
‘Now Menelaus of the great war cry captured Adrastus
alive: for his two horses bolting over the level land
got entangled in a tamarisk growth, and shattered the curving
chariot at the tip of the pole; so they broken free went
on toward the city, where many beside stampeded in terror.
But Adrastus was whirled beside the wheel from the chariot
headlong into the dust on his face; and the son of Atreus,
Menelaus, with the far-shadowed spear in his hand, stood over him.’

In this example, the amount of text hosted by γάρ is longer than in (2) to (5); only
the four events A, B, C, and D t a k e n to g e th e r explain how it came about that Adrastus
was taken alive (E). Here we are in fact close to the Herodotean example with which I
4 For the concept o f character language, i.e., words which are used mainly by characters (in
speeches or passages of embedded focalization), see De Jong (forthc.); for άτιμάω see Griffin
(1986: 45).
5 Cf. οΰνεκα ta ll. 11.21, 54, 79, 627; 13.461; 18.247; 19.45; 20.42, 409; 23.386; 24.607; Od.
3.53; 7.10; 8.200,310; 18.282; 21.367,
6 Other examples: II. 4.467; 6.130 (ουδέ γάρ ούδέ); 14.315; 19.95 (και γάρ); Od. 1.260; 3.262,
276; 4.677; 14.244, 317; 19.186 (κάι γάρ). Cf. also Od. 17.419, where και γάρ starts off a nar­
rative at the beginning.
 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES 179

started and which, wc now note, displays the same structure: the narrator announces
that Pisistratus was tyrant of Athens, then in one step goes back in time to the
moment his father received an omen about his son and in chronological order
recounts how Pisistratus—not without difficulty—acquired the tyranny.
I conclude that the use of γάρ to introduce narratives originated in a typical
archaic (oral?) form of narration, viz. announcing an event and then going back in
time and filling in the details as to how this event came about. When the narrator
goes back in time step by step, γάρ hosts little pieces of narrative information; when
he goes back in a single step, the particle hosts a continuous narrative, sometimes of
considerable length.

Pindar

Most Pindaric odes contain a myth, and a number of these myths are introduced by
γάρ.7 An example is:

(7) άμφί δ' ανθρώπων φρασίν άμπλακίαι

άναρίθΙμητοι κρέμανται- τούτο δ' άμάχανον εύρεΐν,
ö u νυν έν καί τελευτή φέρτατον άνδρΐ τυχεΐν.
κ α ί γάρ Άλκμήνας κασίγνητον νόθον
σκάπτω θενών
σκληρας έλαίας έκτανεν Τί-
ρυνθι Λικύμνιον ( Ο . 7.24-9)8
‘Yet over men’s minds
the clouds of countless errors hover—
it is impossible to know what happens
for the best, now, or in the end.
In Tiryns once, this city’s founder [Tlepolemus],
furious, with a club of gnarled olive wood
battered to death
Alcmena’s bastard brother Licymnius ...’^

K ai γάρ introduces one of three myths which together present a history of Rhodes
in three regressive movements. Pindar returns to his starting point only in 77-8:
‘Here [on Rhodes] Tlepolemus, fleeing Tiryns, found redemption from his bitter
sorrow.’ Γάρ fulfils its explanatory function, in that the myth it introduces explains
how sometimes ‘it is impossible to know what happens for the best’ (the force of

7 Another, more frequent, connector is the relative pronoun, see Des Places (1947:48-50).
8 Other examples: P . 3.26; 5.83; 7.24; N . 9.13; 10.60. The list given in Slater's lexicon (1969:
101-2) under (c) ‘introduces narrative in elaboration of what precedes’ contains quite a number of
instances (0. 4.10; P . 2.25; 3.25; 4.102, 159; 256, 281; 6.32; 8.35; 9.114; N . 7.24; 8.9; 10.65; /.
1.17) where, in my view, γάρ either does not introduce a na rra tive (but a different type of dis­
course) or does not introduce a narrative (but occurs after the narrative has started).
9 1quote here and in (8) the translation ofFJ. Nisetich, P in d a r's Victory Songs. Ballimore 1980.
 IRENE J.F . DE JONG

καί being to signal that what follows is an example which illustrates the point which
Pindar wants to make):101Tlepolemus started by killing his uncle, but in the end his
misfortune turned him into the founder of Rhodes; in other words, it all happened for
the best. The same applies to the other two myths (dealing with the Heliadae and
Helius): they describe an act that seemed harmful at the time, but led to good in the
end.
A somewhat different example is found in:

(8) άπό 5' αυτόν έγώ Μοίσαισι δώσω

καί τό πάγχρυσον νάκος κριού' μετά γάρ
κείνο πλευσάντων Μινυαν, Θεόπομ­
πο! σφισιν τιμαί φύτευθεν.
τίςγ ά ρ άρχά δέξατο ναυτιλίας...; (Ρ. 4.68-70)
O f him [Arcesilas], then, will I sing, and of the Golden Fleece,
for when the Minyan chiefs set sail to fetch it,
the gods sowed honor for his race and him.
What, then, started them on that voyage...?’
(there follows the story of the quest for the Golden Fleece until line 246)

Braswell (1988: a d l o c .) comments: ‘This use of explanatory γάρ is regularly found

‘after an expression denoting the giving or receiving of information, or conveying a
summons of attention’ (Denniston G r e e k P a r t i c l e s 59). It is often used to introduce
a story or myth.’11 Here, in my view, we have a second explanation for the use of
γάρ at the opening of narratives: this time it is not the o u t c o m e o f the narrative which
is announced, the actual story then following in a section introduced by γάρ, but th e
n a r r a tiv e i t s e l f (‘I will sing of the quest for the Golden Fleece’).

Sophocles

Sophocles’ seven extant tragedies contain eight narratives by messengers, reporting

events which have taken place offstage.12 Four of these narratives are introduced by
γάρ, e.g.:

(9) (Creon has asked the messenger how Iocasta died)

Αγ. αυτή προς αύτης. των δέ πραχθέντων τα μέν
άλγιστ’ άπεστιν ή γάρ όψις οΰ πάρα,
όμως δ’, όσον γε κάν έμοί μνήμης ένι,
πευση τα κείνης άθλιας παθήματα.

10 Cf. καί γάρ in Π. 19.95 (and Willcock 1984: ad loc. ‘This combination of particles is regularly
used to introduce a mythological or other example, as illustration of the point that is being
made.’); 24.602; and Od. 17.419.
11 Of the other instances adduced by Braswell (P . 4.14,263, PI. Prt. 320c, A. Pers. 255, Ag. 267,
and S. Tr. 475), A. Pers. 255 and Ag. 267 do not introduce a story or myth.
12 For the corpus, see De Jong (1991: vii, note 5).
 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES 181

όπως γ ά ρ όργη χρωμένη παρήλθ’ έσω

θυρώνος, ϊετ ευθύ προς τα νυμφικά
λ έ χ η ,... (ΟΓ 1237-42)13
‘By her own hand. The pain in what has happened is not for you; for you did
not behold it with your own eyes. Nevertheless, so far as my memory goes,
you will learn that unhappy’s woman’s misfortunes. When, frantic, she had
passed within the antechamber, she immediately rushed at her nuptial bed ... ’

This tendency of Sophocles to begin his messengers’ narratives with a γάρ-clause

differs from Euripides’ habit of opening with an έπεί-clause, e.g.1314

(10) έ π ε ί θεράπνας τησδε Θηβαίας χθονός

λιπόντες έξέβημεν 'Ασωπού ροάς,
λέπας Κιθαιρόνειον είσεβάλλομεν ( Β α . 1043-5)
When we had left the settlements of this Theban land, and gone beyond the
streams of Asopus, we were striking into the uplands of Cithaeron ...

The explanation for this difference is that, while both Sophoclean and Euripidean
messengers begin their narrative in answer to a request for information by their
addressees, Sophoclean messengers tend to precede their narrative by a brief pream­
ble, in which they announce that they are about to recount a (truthful or shocking)
tale, while Euripidean messengers usually embark at once on their tale.15 The use of
γάρ at the opening of Sophoclean narratives, therefore, is another example of the
regular use of this particle after ‘an expression denoting the giving or receiving of
information,’ in (9): ‘you will l e a r n that unhappy’s woman’s misfortunes.’

H e rodotus

As we have seen, Herodotus resembles Homer in his habit of announcing an event or

situation and then retracing his steps, telling the ‘story’ behind this event or situa­
tion16 in a passage introduced by γάρ. I will look at one more example:

13 The other three instances are A). 749; Am. 407; and El. 681. Jcbb nowhere comments on the use
of γάρ; Kamerbeek (1963: ad Aj. 749) notes ‘γάρ in the second verse of a messenger-story also El.
681; cf. O T 1241 (‘be it known’) ’; (1967: ad OT 1241) ‘δπως γάρ: Only here does the
Messenger’s story proper begin, introduced by the typical formula όπως (temporal) γάρ (‘well
then’, ‘eh bicn’). Cf. Tr. 900, OC 1590; E. Med. 1136, Heracl. 800, Andr. 1085, Ion 1122, El. 774,
I T 260, 1327, Mel. 1526, Da. 1043, JA 1543.’ This note is slightly misleading in that γάρ occurs
only in two of these twelve passages (Heracl. 800 and SA 1543), the others are introduced just by
a temporal modifier.
1/1 See Rijksbaron (1976), who also offers a convincing explanation for the rationale behind open­
ings with έπεί.
l5"Four times a Euripidean messenger-speech opens with a prefatory remark; thrice the ensuing
narrative opens with έπεί γάρ (Ale. 158, Heracl. 800, JA 1543). Twice a Sophoclean messenger-
speech opens with έπεί (Jr. 900, OC 1590), and is still preceded by an introductory remark.
16 See Lang (1984:1-7) for what she calls ‘structuring by direction-statement’ in Herodotus.
182 IRENE J .F, DE JONG

(11) Χρόνφ δέ ού πολλώ ύστερον κ α ι Όροίτην Πολυκράτεος τίσ ιες

μετηλθον. μετά γ ά ρ τόν Καμβύσεω θάνατον κα ί των μάγων την
βασιληίην μένων έν τήσι Σάρδισι ό Όροίτης ώφέλεε μέν ούδέν Πέρσας
ύπό Μήδων άπαραιρημένους την άρχήν'(3.126.1)17
‘Not long after, Oroetes, too, was overtaken by the powers that avenged Poly­
crates. After Cambyses had died and the Magians had taken over the king-
ship, Oroetes staying in Sardes in no way helped the Persians to regain the
power taken from them by the Medes.’

There follows the story of how Oroetes estranged himself from the Persians and in
the end was killed by their king Darius. The narrative ends two chapters later with a
typical e p a n a l e p s i s : ούτω δή Όροίτην τόν Πέρσην Πολυκράτεος του Σαμίου
τίσιες μετηλθον (3.128.5), ‘in that way the powers that avenged Polycrates, the
Samian, overtook Oroetes the Persian.’

Lysias

A standard element in Lysias’ speeches is the n a r r a t i o , the presentation of the cir­

cumstances of the—alleged—crime. The majority of these n a r r a tio n e s in the C o r p u s
open with γάρ, e.g. :
L y s ia c u m

(12) έγώ τοίνυν έξ αρχής ΰμιν άπαντα έπιδείξω τα έμαυτού πράγματα, ούδέν
παραλείπων, άλλα λέγων τάληθίγ...
Έγώ γ ά ρ , ώ ’Αθηναίοι, έπειδή εδοξέ μοι γημαι και γυναίκα ήγαγόμην
εις την οικίαν, τόν μέν άλλον χρόνον οϋτω διεκείμην... (1.5-6)18
‘I shall therefore set out for you my whole story from the beginning, leaving
out nothing but telling the truth ...
When I, Athenians, decided to marry and brought a wife into my house, for
some time I was disposed ...’

Sicking (1993: 20) gives the following analysis of this instance of γάρ, which is in
line with the argument of this paper: ‘In 1.6 γάρ marks the beginning of the narra­
tion, announced by the speaker in 5, which is only concluded with ούτως in 27, and
so in its entirety falls within the scope of the particle: by means of the information
which he presents in the form of a narrative the speaker acquits himself of a promise
made in 5.’ In other words, the use of γάρ here again belongs to the larger subcate-

17 Other examples: 1.1.1, 65.1, 68.2; 2.118.2 (the naiTativc announced by τάδε); 3.34.1 (τάδε),
50.1; 4.78.1, 52.1; 6.52.1, 126.1; 7.119.1 (τον λόγον), 135 (τάδε), 147.2 (τη ... άλλη), 148.2
(ώδε), 239.2; 8.87.2 (τάδε), 105.1,116.2,120.2 (τάδε). Of course, there are numerous instances
where γάρ, as in Homer, introduces brief pieces of information from the past; these instances have
been left out of account
18 Other examples: 2.4, 3.6, 6.21, 7.4, 13.5,14.25, 16.4,19.12, 21.1, 22.2, 23.2. For narrationes
which do not open with γάρ, see: 12.4,17.2,32.4.
 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES 183

gory of γάρ following ‘after an expression denoting the giving or receiving of

information.’

Varia

In this last section I present two final—and particularly fine—instances of γάρ at the
opening of a narrative, from different authors:

(13) δοκεΐ τοίνυν μοι, έφη, χαριέστερον είναι, μΰθον ύμΐν λέγειν.
Ή ν γάρ ποτέ χρόνος δτε θεοί μεν ήσαν, Θνητά δέ γένη ούκ ήν
‘It seems to me to be more elegant to tell a story. There was a time when the
gods existed, but the race of mortals not.’ (PI. P r t . 320c)

We find ourselves here at the beginning of an (argumentative) myth told by Protago­

ras (320c-322d). The formulation recalls the O nce upon a time ...’ with which
fairytales tend to open.19

(14) Άρχόμενος σέο Φοίβε παλαιγενέων κλέα φωτών

μνήσομαι οϊ Πόντοιο κατά στόμα και διά πέτρας
Κυανέας βασιλήος έφημοσυνη Πελίαο
χρΰσειον μετά κώας έύζυγον ήλασαν ’Αργώ.
Τοίην γάρ Πελίης φάτιν έκλυεν, ώς μιν όπίσσω
μοίρα μένει στυγερή, τοΰδ1άνέρος, δντιν’ ϊδοιτο
δημόθεν οίοπέδιλον ΰπ’ έννεσίησι δαμήναι (Α.Ρ. A r g o n a u t i c a 1.1-7)2021
‘Taking my start from you, Phoebus, I shall recall the glorious deeds of men
long ago who propelled the well-benched A r g o through the mouth of the
Pontus and between the Dark Rocks to gain the golden fleece <at the behest
of King Peliasx For such was the oracle which Pelias had received, that a
hateful fate awaited him in the future—destruction caused by a country man
whom he would see wearing only one sandal.’2·1

Having announced the subject of his poem (1-4), the narrator launches his actual nar­
rative with γάρ. It can be said to be triggered by the announcement that he will
recount a story ( Ί shall recall the glorious deeds’), but it also sets out to explain why
Pelias had come to organize the quest for the golden fleece. Once more, it will not do
to take only the γάρ-clause itself as the explanation of what precedes; why should
Pelias’ oracle lead to an expedition by the Argonauts? To find out, we have to read
on:

19 Other examples: Ti. 21a; X. Smp. 1.2.

20 Other examples: 2.476; 4.529,988, and 1436.
2 1 1 quote the translation of R. Hunter, Jason and the Golden Fleece (The Argonautica). Oxford
1993.1 have added ‘at the behest of Pelias’ (έφημοσύνη Πελίαο: 3), which, by mistake, is miss­
ing from his translation.
 IRENE J.F . DE JONG

‘Not long after, in accordance with your prophecy, as Jason was crossing on
foot the streams of the Anaurus in winter, he saved one sandal from the mud
but the other was caught by the current and he abandoned it to the depths. He
hurried straight on to Pelias’ palace, to take part in a sacred feast, which the
king was offering to his father Poseidon and to all the other gods-except
Pelasgian Hera to whom he paid no regard. As soon as Pelias saw Jason he
realized, and devised for him the challenge of a voyage which would be full of
suffering, so that either on the sea or among a foreign people he might lose
all chance of safe return.’ (8-17)
In line 17 the narrator has returned to his point of departure, the voyage o f the
Argonauts, and now we understand the relation between that expedition and the ora­
cle: by sending Jason on this dangerous mission, Pelias in fact hopes to get rid of
them (and hence to prevent the oracle from being fulfilled).

Conclusion

In this paper I have argued that one of the uses of explanatory γάρ—one not noted
by Denniston—is to introduce embedded narratives. The fact that γάρ fulfils this
function is not surprising. Γάρ is a typical ‘push’ particle, to use the terminology
introduced elsewhere in this volume by S. Slings. As such, it introduces background
information, which sometimes takes the form of a narrative. When used in this way
(hosting a piece of text larger than one sentence), γάρ is often left untranslated or
acquires a meaning like ‘now you should know’, the French ‘eh bierì , or the Dutch
‘moet je weten’.
This use seems to have originated in a typically archaic manner of storytelling,
which consists of announcing the outcome of an event or the event itself, and then
going back in time and filling in the details of how this event came about. In the sec­
ond place, embedded narratives are often announced as such ( Ί will now tell a ...
story’); here we are dealing with the common use of γάρ following after ‘an expres­
sion denoting the giving or receiving of information. ’ It may be instructive to note
that in Latin, too, the particle nam is regularly used to introduce embedded narra-
dves.22

Bibliography

Braswell, B.K.
1988 A Commentary on the Fourth Pythian Ode o f Pindar. Berlin-New York
Denniston, J.D.
19542 The GreekParticles. Oxford
Griffin, J.
22 See Kroon (1995: 148-9).
 ΓΑΡ INTRODUCING EMBEDDED NARRATIVES 185

1986 ‘HomericWords and Speakers’. JHS 106,36-57

Jong, I.J.F. de
1991 Narrative in Drama. The Art o f the Euripidean Messenger-Speech. Leiden
forthc. ‘Narrator Language Versus Character Language: Some Further Obser­
vations’. In: F. Létoublon (éd.) Colloquium Milman Parry. Amsterdam
Kamerbeek, J.C.
1963 The Plays o f Sophocles. Commentaries I, The Ajax. Leiden
1967 The Plays o f Sophocles. Commentaries IV, The Oedipus Tyrannus. Leiden '
Kroon, C.
1995 Discourse Particles in Latin. A Study o f nam, enim, autem, vero and at.
Amsterdam
Lang, M.L.
1984 Herodotean Narrative and Discourse. Cambridge (Mass.)-London
Places, E. des
1947 Le pronom chez Pindare. Recherches philologiques et critiques. Paris
Rijksbaron, A.
1976 ‘How Does a Messenger Begin His Speech? Some Observations on the
Opening-Lines of Euripidean Messenger-Speeches.’ In: J.M. Bremer, S.
Radt, C J . Ruijgh (eds), Miscellanea tragica in honorem J.C. Kamerbeek.
Amsterdam, 293-308
Schadewaldt, W.
1938 Jliasstudien. Leipzig
Sicking, C.M J . & J.M. van Ophuijsen
1993 Two Studies in Attic Particle Usage. Lysias & Plato. Leiden
Slater, W.J.
1969 Lexicon to Pindar. Berlin
1983 ‘Lyric Narrative: Structure and Principle.’CA 2,117-32
Willcock, M.M.
1984 The Iliad of Homer, II. London
 ADVERB OR CONNECTOR?
The Case o f καί ... δέ

Al b e r t r i j k s b a r o n
Universiteit van Amsterdam

1. Introduction

In the general introduction I made mention of Denniston’s carelessness in syntactic

matters, using his discussion of άλλα γάρ as an example. Another particle combina­
tion illustrating this attitude is καί ... δέ. In fact, the treatment of this combination
shows us Denniston at his most easy-going. On p. 199 he says: ‘This is a natural
enough combination, the former particle denoting that something is added, the latter
that what is added is distinct from what precedes.’1 As appears from a footnote this
analysis was inspired by Jebb: ‘Jebb, on S. P h . 1362, argues for the view that, in καί
... δέ, κ α ί is the conjunction, while δέ means “on the other hand”, “also”’, on
which Denniston comments: ‘This is, I think, the right explanation of most of the
passages. But there are others (see (2) below) in which δέ seems to be the conjunc­
tion, while κ α ί means “also”. ’12 One immediately wonders, first, how Denniston can
advocate a translation of κ α ι ... δέίη which δέ, which should express ‘distinctness’,
means ‘also’, while at the same time admitting that there are cases where it is καί that
means ‘also’. Rather surprisingly, the net result of his analysis is that in either case
the combination κ α ί... δέ means ‘and also’. Second, one also wonders in what way
the combination of καί and δέ differs from that of δέ and καί. Most importantly,
Denniston’s remarks on δέ amount to saying that in most cases of καί ... δέ it is δέ
which is the adverb, although the term as such is avoided. But thereby he flatly con­
tradicts his own rather apodictic statement on p. 162: ‘Except in the apodotic use, δέ
is always a connective.’ Matters are further complicated by Denniston’s treatment of
1 Incidentally, in his discussion o f δέ Denniston uses the terms ‘contrast’, 'adversative' (p. 162)
and ‘opposed’ (p. 165), but not ‘distinct’; these terms are considered, moreover, values of the con­
nective use only (p. 162).
2 Jebb is less categorical than Denniston suggests, for his remark runs: ‘The usual account of it is
that the κα ί = ‘also’, while δέ = ‘and'. This suits those instances in which, as here, κ α ί ... δέ is
preceded by a full stop, or by a pause’ (I will come back to the Philoctetes passage below - AR)
‘but it is less natural where κ α ί... δέ links a new clause to a preceding one m the same sentence;
as in Thuc. 4.24: id. 9.71 (read: 6.71).’ (On 6.71 see.further below - AR.) Jcbb’s appeal to punctua­
tion to decide this matter involves, of course, a circular argument, there being no way to punctuate
independently from our syntactic analysis.
 ALBERT RJJKSBARON

ούδε ... δέ, the negative counterpart of κ α ι... δέ. On p. 203 he remarks: ‘Here δέ is
clearly the connective, and ούδε is adverbial. ’ Why this is so remains unexplained.
Denniston might also have turned to Kiihner-Gerth for support. They mention
our particles on page 2, 253, significantly n o t in the section on ‘καί, e t i a m , als
Adverb’, and say that epic και δέ (with no intervening word) = ‘und andererseits,
hinwiederum, ferner, und doch’, implying that δέ is adverbial. Rather confusingly,
however, they speak some lines further about: ‘Die Verbindung der ... Konjunktio­
nen (plural) καί ... δέ’, as if both panicles perform the same function.3 But further
on, in the section on δέ, they emphatically state, page 2, 261, that δέ originally was an
adverb ‘mit der Bedeutung a n d e r e r s e i t s , d a g e g e n ’, and that this meaning is
‘ganz deutlich’ preserved in Homeric καί δέ and post-Homeric καί ... δέ, as well as
in their negative counterpart ουδέ ... δέ.
To conclude these preliminary remarks I should add that while Kiihner-Gerth
categorically, and Denniston with some reservations, defend the view that, of καί ...
δέ, it is δέ which is the adverb, many commentators, as we will see, categorically take
the opposite position.

For reasons that will become clear shortly I have arranged my examples into two
groups. First, I shall discuss the use of καί ... δέ as a connective device, to use a
neutral term, between single words (adjectives, nouns, and verbs). For practical rea­
sons I have further divided the group of nominal constituents into two sub-classes,
according to the number of the words involved. In the second part of my paper the
use of καί ... δέ to connect clauses and sentences will be discussed.4*

2. T h e use o f κ α ί ... δ έ connecting single w ords3

2.1. κ α ί ... δ έ connects tw o items

3 LSJ .?v δέ II 2 b note, not very helpfully: ‘In the combination καί δέ (-), κ α ί... δέ (-), each par­
ticle retains its force.’ Schwyzcr-Debmnner do not discuss καί (...) δέ.
4 I concentrate on Attic Greek and Herodotus, but will refer to some Homeric instances in fn. 30.
Καί ... δέ is distributed rather unevenly; also, not seldom there is ms. variation or δέ has been
removed per contecturam. The numbers for some Attic writers arc (fragments not included): Aeschy­
lus: 9; Sophocles: 1 or 2 (at Ant. 432, a’s χήμεις ίδόντες δ' may conceal χήμείς δ' ίδόντες):
Euripides: 4; Aristophanes: 3, all in Pax; Thucydides: 6; Plato: 35; Lysias: 2; Xenophon: approx.
265; Demosthenes: 22; Theophrastus’ Characters approx. 60. For Herodotus the numbers are: 6; for
Homer: 33. Homer has only the combination καί δέ, while in Herodotus and Attic Greek only καί
... δέ occurs. There is nothing exceptional about this. Similar discrepancies are found with particle
combinations like καί γάρ (the only combination occurring in Homer), as against Attic καί γάρ
and καί ... γάρ. It is a priori not clear whether these formal differences reflect a possible semantic
or pragmatic difference. The uses of particles that are known to both Homer and Attic Greek may, of
course, vary considerably, some examples being περ, κ α ί μήν, μένιοι and οΰν. In principle, then,
the fact that Attic Greek uses κ α ί... δέ in one way does not exclude that Homeric καί δέ may have
been used in another way. But let me add immediately that I do not think this is the case. See fn.
30.
3 This use is not known to Homer, who uses καί δέ only at clause and sentence level.
 ADVERB OR CONNECTOR? 189

(1) "Αρξομαι δέ άπό των προγόνων πρώτον δίκαιον γάρ αύτοΐς και πρέπον
δέ άμα ... την τιμήν ταύτην της μνήμης δίδοσθαι
(‘I shall speak first of our ancestors for it is right and at the same time fitting
... to give them this place of honour’, Th. 2.36.1; unless indicated other­
wise,the translations are those of the Loeb-series, sometimes with adaptations
to bring out the value of δέ)

The words connected by κ α ί.. δέ are two predicative neuter adjectives.

Classen-Steup ad loc. render: ‘a b e r auch zugleich ... geziemend’, referring to
1.132.4. There they had, in fact, observed: ‘In der Verbindung der Partikeln καί .. δέ
(2, 36, 1; 4, 24, 2; 6, 71, 2; 7, 56, 3 ... u. 8, 67, 3) ist καί nicht Kopula, sondern
a u c h . ’ But this is very unlikely. The point is that if καί in (1) is taken as an addi­
tive adverb, and thus as optional,6 δέ must be capable of connecting the two adjectives
all on its own. This, however, is not the case, as appears from the unacceptability of
ex. (2), both with δέ as ‘and’ and as ‘but’:7

(2) *δίκαιον πρέπον δέ

On the other hand, καί in (1) is, of course, capable of connecting the two adjectives
on its own, which strongly suggests that it is δέ that is an (optional) adverb here. To
be sure, δέ may connect two adjectives, but only as a ‘balancing adversative’
(Denniston p. 165), v iz . when the second adjective contrasts with an implication of
the first adjective, as in (3) and (4).8

(3) πονηρά γ', ώ Νεφέλαι, δίκαια δέ

(‘Ah Clouds, hard words, but they’re fair ...’, Ar. N u . 1462; transl. Sommer­
stein)

(4) οϋτω δή τά μέν στοιχεία άλογα καί άγνωστα είναι, αισθητά δέ-
(‘Thus the elements are not objects of reason or of knowledge, but only of
perception’, PI. T h t. 202b)

Otherwise, the connection must be established by καί, as in (5):

6 Cf. Engl. John, Peter and (also) Richardland Rickard (too). For optionality as a defining charac­
teristic of most adverbs see Pinkster (1972: 75-6).
7 Strangely enough, this fundamental difference between καί and δέ, viz. that the latter in principle
is not suited to connect single words, is mentioned more or less in passing by Denniston, in one
sentence and a footnote on p. 162.
8 Such implications arc mostly a matter of convention. They may be conventionalized to such a
degree that they are virtually an inherent semantic feature of the adjective involved. Thus, poor
always implies ‘not-happy’, as may be inferred from the fact that poor but happy is a well-formed
expression, while *poor but unhappy is not. However, they may also be of a more informal, indi­
vidualized, nature, as when someone says This wall is blue but (and yet) beautiful. Apparently for
this particular speaker ‘blueness’ implies ‘ugliness'.
 ALBERT RUKSBARON

(5) ταΰτα γάρ δίκαια καί πρέποντα άμα τηδε τη πόλει άποκρίνασθαι
(‘This answer is just and at the same time consistent with the dignity of the
city’, Th. 1.144.2)9

Now it might be objected that it is precisely the presence of καί which makes the
otherwise unacceptable δέ in (2) acceptable, on the analogy of e.g. English and
Dutch. In these languages combinations like that of (2) are equally unacceptable:
j u s t i f i e d b u t f i t t i n g , * g e r e c h tv a a r d ig d m a a r g e p a s t; interestingly, however, the pres­
ence of a ls o , and o o k turns them into fully acceptable ones: i t is j u s t i f i e d , b u t a l s o f i t ­
t in g , h e t is g e r e c h tv a a r d ig d , m a a r o o k g e p a s t . These combinations probably counter
a possible objection like O n l y j u s t i f i e d ? , A l l e e n m a a r g e r e c h t v a a r d i g d ? ; b u t and
m a a r may be considered ‘abbreviations’, so to speak, of n o t o n l y .. ., b u t a l s o , n i e t
a l l e e n .... m a a r o o k . Likewise the addition of adverbial καί might make the use of δέ
to connect non-contrastive items acceptable. Some commentators explain καί ... δέ,
in fact, in this way. Thus, Frohberger remarks, in his A n h a n g to Lys. 19.5: ‘Die
Verbindung ... καί ... δέ dient stets der gradatio und betont den eingeschobenen
Begriff (hier = ού μόνον έγώ ... άλλα καί ύμάς).’ However, we need not, and
indeed cannot, follow this way out. First, if the n o t o n l y ... b u t α/ίο-approach were
correct (i.e. not just as an explanatory but as a grammatical solution) one would
expect άλλα κ α ί rather than καί... δέ. Second, there sometimes may be gradatio
involved, but this is definitely not ‘stets’ the case.910 Third, and most important, there
are examples of κ α ί ... δέ where καί without any doubt is the connector, like:

(6) άλλ' α ίνίττετα ι... καί οΰτος κ α ί άλλοι δέ πονηταί σχεδόν τι πάντες
(‘But he......like almost every other poet, speaks in riddles’, [PL] A l e . II
147b8) (δλλοι ci. Bumet, άλλοι B : οί άλλοι T)

Of course, the presence of preparative καί here is of prime importance for our anal­
ysis: we must be dealing with corresponsive καί .... καί, which means that, of καί ...
δέ, it is καί which is the connector. Another example of two nominal elements con­
nected by καί .... καί ... δέ is X. S m p . 2.9 (έν πολλοίς μέν ... καί άλλοις δήλον
καί έν οΐς δ’ ή πάίς ποιεί κτλ). Often, the two nouns are part of a series of nouns
9 Scholars often mention cases where κ α ί seemingly connects items that are semantically opposed,
e.g. E. M e d . 1243 τα δεινά κάναγκάία ... κακά. In a small but (and?) insightful monograph,
Blomqvist (1979) convincingly argues that this fact should not induce us to assume the existence
of a ‘καί adversativum’. An adversative particle is ‘eine Partikel, die nicht nur Gegensätze
verknüpft, sondern diese auch als Gegensätze bezeichnet’ (p. 56); cf. also Slings (1980: 122) ‘... an
adversative interpretation (of καί) can be suggested by the semantic relations of two members coor­
dinated by κα ί’. Regrettably, Blomqvist does not discuss κα ί (...) δέ. Incidentally, I do not think
that in a case like M e d . 1243 there is a semantic opposition. By the use o f καί the speaker/author
deliberately presents the two qualifications as being valid at the same time: for the speaker there is
no contrast between them. Cf. an example like Engl. This fo o d is very unhealthy a n d very tasteful,
where the standard implications of unhealthy make us expect but very tasteful. See also fn. 8.
10 Thus, there is clearly no gTadatio in the very example used by Frohberger, which runs: ακούω
γάρ έγωγε, κα ί ύμών δέ τούς πολλούς ονμαι είδέναι, ότι κτλ. The κ α ί-clause (which belongs
to my second type, see below § 3) is a parenthesis rather than a highlighting clause. Also, the pres­
ence of -γε with έγώ indicates that the speaker is primarily concerned with himself.
 ADVERB OR CONNECTOR? 191

connected by καί (see § 2.2 below), e.g. πελταστάς πολλούς καί τοξότας καί
σφενδονήτας καί Ιππέας δε ... in ex. (15).
But what, then, is the function of δέ in καί ... δέ? Like Denniston, and like K-G,
for that matter, I think that this particle presents the item concerned as d i s t i n c t from
the first item, not semantically, however, as we have seen, but pragmatically.11 That
is, while κα ί expresses the idea that the two items semantically belong together (cf.
fn. 11), δέ indicates that the (referent of the) second item should be considered in its
own right, and is, thus, at least as important as the first item. I may perhaps quote here
the words of Caroline Kroon in her admirable study of a number of Latin particles,
on the very similar Latin combination e t ... e t ... a u t e m : ‘The expression e t ... e t ...
a u t e m : is used to emphasize on the one hand the coordination of corresponding units
of information, and on the other the individuality and mutual discreteness of the
coordinated elements’ (Kroon 1995: 231).1112 On p. 229 she had argued that by the
presence of a u t e m the ‘focus of attention shifts from A to B’. The eventual effect of
δέ in a καί ... δέ group may be rendered in various ways, e.g. by ‘and on the other
hand’, ‘and apart from that’, ‘and again’, ‘and further’. Often the effect would seem
to be the same as the effect conveyed by English ‘for that matter’, Dutch ‘trouwens,
overigens’ (= ‘apart from’): it anticipates a possible question on the part of the
addressee ‘what about item X?’, and indicates that the speaker is aware that that item
is relevant as well.
On the basis of this discussion I conclude that in (1), too, it is καί which is the
connector. As suggested above, the function of δέ is to individualize the second item;
it is perhaps best rendered by ‘for that matter’ or ‘on the other hand’.13
11 Expressing 'distinctness' in the pragmatic sense may be the common function o f all uses of δέ:
in κ α ί ... δέ, in a p o d o si, in combinations like ό μέν ... ό δέ, and at clause and sentence level. In
the latter use δέ is (syntactically) also a connector, with the semantic function ‘adds y to x ’. By set­
ting off the unit of information it introduces from the previous information unit, δέ often not just
adds information but implies that the two units contrast q u a content. This, in turn, often correlates
with a shift in Topic or Theme. The difference with και is perhaps that καί only connects and adds,
and does not set off: the two units should be taken as forming together one non-contrastive informa­
tion unit. Asyndeton, on the other hand, does not connect, but only sets off: the unit o f information
is presented as having no formal links at all with what precedes. Thereby this information gains a
certain independence vis-à-vis the information that precedes; the addressee is invited not to look
backward, and to concentrate on the unit of information at hand. In practice, this means, especially in
a narrative, that asyndeton signals a point of special interest, often marking the beginning of a new
piece of narrative. Cf. Rijksbaron (1993). For the function of δέ see also Bakker (1993) and Sick­
ing & Van Ophuijsen (1993: Ilf., 47).
12 Kroon calls this ‘mixture o f (formal) conjunction and (semantic or pragmatic) disjunction’
‘remarkable’; for κ α ί... δέ, too, this qualification would seem more appropriate than Denniston’s
‘this is a natural enough combination’. In formal pragmatic terms one might say that in e t ... autem
the two particles operate at different levels of discourse, et at the representational, and a u tem at the
presentational level, cf. Kroon (1995: 69 and 226ff.; for these distinctions cf. also Wakker (this
volume p. 231)). The same analysis applies to καί ... δέ. I should perhaps mention here the possi­
bility that κα ί ... δέ, with adverbial δέ, fills the gap, so to speak, created by the impossiblility of
combining κα ί ‘and’ and καί ‘also’: *καί καί.
13 This pragmatic function correlates with the presence o f a syntactic boundary before the item fol­
lowed by δέ (in (1): πρέπον), a boundary which is due to δέ being a postpositive. For the function
of Ösand other postpositives as boundary markers sccRuijgh (1996: 629ff.), Dik (1995: 35ff.). As
for the relationship between this use of δέ, and that of αδ ‘on the other hand, further’; see below p.
195.
 ALBERT RIJKSBARON

Another example of κ α ι... δέ connecting predicative adjectives is:

(7) το μέν οΰν μή ούχι ήδέα είναι τα ήδέα λόγος ούδείς αμφισβητεί- κακά δ’
οντα αυτών τά πολλά καί αγαθά δέ, ώς ήμεΐς φαμέν, όμως πάντα σύ
προσαγορεύεις αγαθά α ύτά ,...
(‘Now of course nobody attempts to maintain the thesis that pleasant things
are not pleasant; but though they are in some cases (indeed in most) bad and
in others good ... nevertheless you designate them alias good, ...’, PI. P h i l .
13b, transi. Hackforth)

Again, δέ cannot possibly be taken as the connector, for the same reasons as those
set out above. I am not sure about the eventual effect of δέ in καί άγαθά δέ; perhaps
the addition serves, after the information that most pleasant things are bad, as a kind
of afterthought: for the sake of the argument Socrates concedes explicitly that pleas­
ant things are ‘on the other hand’ sometimes good, ‘sometimes’ being an effect of
the presence of τα πολλά.

The next three examples all conform to the pattem set out above: καί cannot be
omitted, and δέ has an ‘on the other hand’ or ‘for that matter’ value.

(8) αύτίκα γάρ έδόκουν οί "Ελληνες καί πάντες δέ άτάκτοις σφίσιν έπιπε-
σείσθαι
(‘for the thought of the Greeks, and of all the rest in fact, was that he would
fall upon them immediately, while they were in disorder’, X. A n . 1.8.2)

(9) εί μέντοι ρήτωρ έστίν οίους ένίους των λεγόντων έγώ καί υμείς δ’ όρατε
... ούκ αν είην οΰτος έγώ
(‘But if by orator he means one of those speakers such as I and you, for that
matter, often see,... I cannot be one’, D. 21.189; ‘for that matter’ added)14

(10) ωμήν ... προσηκειν Εύβουλίδη, καί πάσιν δ' ό σ ο ι... κατηγοροΰσιν, ...
(‘I should have thought... that it was fitting for Eubulides and for all those, for
that matter, who are ... making accusations ...’, D. 57.4; ‘for that matter’
added)15

καί ... δέ may also connect two predicates, cf.:

14 In their index to N eun Philippische Reden, Leipzig 1886, s.v. κα ί ... δέ, Rehdanl/.-BIass have
collected a great number of instances from Demosthenes. Like Frohberger (see above), they lake
them all as ‘aber auch’, ‘mit Betonung des dazwischen stehenden W ortes’. Strangely enough,
MacDowell in his commentary on the in Meidiam not even mentions the difficulties involved in
κα ί ... δέ.
15 Other examples arc: έκπ ίπτει ύ Άβραδάτας καί άλλοι δέ ... (‘Abradatas and others were
thrown to the ground...’, X. Cyr. 7.1.32), ώ Ύστάσπα, καί o i άλλοι δέ οίπαρόντες (‘Hystaspes
and you others here’.X . Cyr. 8.4.17); PI. Leg. 862c.
 ADVERB OR CONNECTOR? 193

(11) έγώ νή Δί’ έρώ, καί γράψω δέ, ώστ’ αν βούλησθε χειροτονήσετε.
( ‘I solemnly promise that I will answer it and will also move a resolution, for
which you can vote, if so disposed’, D. 9.70)

Again, connection by just δέ is not possible: *έρώ γράψω δέ, while the sequence
έρώ καί γράψω is perfectly viable. This means once more that δέ is optional and
should be taken as an adverb. Both ‘for that matter’ and ‘moreover’ would seem
suitable ‘translations’ to make clear its pragmatic function.

2.2. κ α ί ... δ έ connects two items in a series of three or more items

The examples of this section do not differ basically from those discussed above, but
the connective function of καί is even clearer, since the other items are usually pre­
ceded by καί as well. The καί ... δέ-phrase is always the last item of the series. Let
me repeat that in none of these cases καί can be omitted. In all cases the function of
δέ is to set off the item it modifies from the previous items.

(12) ... π ερί τά υγιεινά καί νοσώδη έθελήσαι αν φάναι μή παν γύναιον καί
παιδίον, καί θηρίον δέ, ικανόν είναι ία σ θ α ι...
(‘.,. he might be willing to say that in matters of health and disease not every
woman or child—or beast, for that matter—is able to cure itself’, PI. T h t.
171e; ‘for that matter’ is present in the Loeb-translation)

Since we are dealing here with a series of ever more unlikely items, the καί θηρίον
δέ phrase being the last in the series, the status of this phrase can perhaps also be
indicated by rendering ‘let alone a beast’, the ‘let alone’-effect being due to the pres­
ence of the negative with the first item.

(13) πρώτος δέ είσεισιν ό τό σταδίον άμιλλησόμενος συν τοίς οπλοις, δεύ­

τερος δέ ό τον δίαυλον, καί τρίτος ό τόν έφίππιον, καί δή καί τέταρτος ό
τον δόλιχον, καί πέμπτος δέ όν άφήσομεν πρώτον ώπλισμένον ...
(‘First, then, there shall enter the man who, with his arms, is to run the furlong,
—second, the runner of the quarter-mile,—third, the half-miler,—fourth, the
runner of the three-quarters,— and fifth, the runner in the long-distance race
whom we shall despatch first, fully armed’, PI. L g . 833a-b)

In this example the δέ after δεύτερος sets off all the items that are to follow from the
first one, the σταδιοδρόμος, which had already been given a separate treatment in the
preceding sentence. Within the group set off by δέ, the connection is brought about
by καί, while δέ individualizes the last item. Obviously, there is no question here of a
‘for that matter’ effect, since the last item is the most important for the description
that follows.
 ALBERT RITKSBARON

(14) π α ρ είχε δέ ή θεός άλφιτα....... κ α ί των θυομένων ... λόχος, καί

θηρευομένων δέ. καίγόρ Θήραν έποιούντο ...
(‘And the goddess would provide for the banqueters barley m eal... and a por­
tion of the sacrificai victims ... as well as of victims taken in the chase. For
(they) used to have a hunting expedition X. A n . 5.3.9)

Both των θυομένων and θηρευομένων are partitive genitives with λόχος, connected
by καί.

(15) ... όρώντι δέ πελταστός πολλούς καί τοξότας κ α ί σφενδονήτας καί

ιππέας δέ ...
(‘... (his) eyes rested upon ... and likewise upon a gTeat body of peltasts,
bowmen, slingers, and horsemen also X. A n . 5.6.15)

Here, too, the last item is set off from the rest; and naturally so, for the first three
groups all operate on foot.

(16) ... ποιήσειV ... καί άνεμους κα ί ϋδατα καί ώρας καί οτου δ’ αν άλλου
δέωνται των τοιούτων.
(‘(do they imagine) that they will create winds, waters, seasons and such
things to their need ...’, X. M e m . 1.1.15)

‘Setting o ff would also seem to be the function of δέ in the next examples.

(17) ε ί μη ... σαφώς διώρισεν, τί πρώτον δει ποιέίν τούς άντιδεδωκότας καί
τί δεύτερον κ αι τάλλα δ' εφεξής, ούκ οιδ’ ...
(‘For if he had not clearly defined for us what is the first thing to be done by
those who have offered an exchange, and what the second, and so on in due
order.Ido not know [D.] 42.1)

(18) είπείν μέν οΰν μοι ταΰτα προς τον Άρχεβιάδην καί τον Άριστόνουν καί
προς αύτόν δέ τον Κηφεσιάδην ούδέν διαφέρει·
(‘to suggest what you propose to Archebiades and Aristonoiis and to Cephe-
siades himself, for that matter, can cause me no trouble’, [D.] 52.11; ‘for that
matter’ added)

Now syntactically, an example like (18) does not differ from an example like (19),
which exhibits the combination καί ... δή:

(19) σκεψώμεθα δή ... ποιά έστιν α ήμας ώφελει. ύγίεια , φάμεν, καί ισχύς
καί κάλλος κ α ι πλούτος δή·
(‘Then let us see ... what sort of things they are that profit us. Health, let us
say, and strength, and beauty, and wealth ...’, PI. M e n . 87e)
 ADVERB OR CONNECTOR? 195

or from (20), featuring κ α ί... γε:

(20) ετεραι δέ γέ είσι των τεχνών αϊ διά λόγου παν περαίνουσι, καί έργου ...
ή ούδένος προσδέονται ή βραχέος πάνυ, οΐον ή αριθμητική κ α ί λογισ­
τική καί γεωμετρική καί πεττευτική γε καί αλλαι πολλαί τέχναι
(‘But there is another class of arts which achieve their whole purpose through
speech and ... require either no action to aid them, or very little; for example,
numeration, calculation, geometry, diaught-playing, and many other arts PI.
G r g . 450d)

or, again, from καί ... αΰ in:

(21) αλλά μεν δή ή γε αρετή έκαστου, καί σκεύους καί σώματος καί ψυχής
αύ κ α ί ζώου πόντος, ...
(‘But surely the virtue of each thing, whether of an implement or of a body, or
again of a soul or of any live creature,...’, PI. G r g . 506d)

On καί ... δή Denniston remarks, p. 254: ‘Καί ... δή joins sentences, clauses, and
single words’, e.g. ‘at the end of a catalogue’, the passage from M e n o being an
example. And on p. 253 we read: ‘These combinations ( v i z . καί ... δή and κ αί δή
καί) signify that the addition made by καί is an important one. They thus differ from
καί ... γε, which merely stresses the fact that an addition is made.’ One wonders why
Denniston did not include καί ... δέ in his discussion. I believe, incidentally, that the
values of δή and γ ε are different. In a case like (19), δή would seem to express the
idea that the addition of the item it modifies will not come as a surprise to the
addressee.16 Again, γε in καί ... γε rather indicates that to the list which exemplifies
the statement of the main clause the item concerned c e r ta in ly belongs, an effect of the
limitative value: ‘if anything else, then certainly this item’: πεττεία being a game,
there is no έργον whatsoever involved in ή πεττευτική.17
As for κ αί ... αΰ (and καί αΰ), which contains the adverb that expresses p a r
e x c e lle n c e ‘on the other hand, in turn’ and related nuances, one may well ask in what
respect it differs from καί ... δέ. In general, it would seem that αΰ expresses a more
marked contrast than καί ... δέ. Thus, in (21) αΰ contrasts two sets, each consisting
of two items, that arc not just opposed pragmatically but s e m a n t i c a l l y . In fact, καί
(...) αΰ, unlike καί ... δέ, often connects antonyms, e.g. φιλόλογος ... κ α ί αΰ
μισόλογος (PI. L a . 188c), χρήσιμόν τε καί ώφέλιμον καί άχρηστον αΰ καί
βλαβερόν ( R e s p . 518e).18, See also p. 206, καί α ΰ ... ωσαύτως.
16 Cf. Van Ophuijsen’s adaptation of Brugmann-Thumb’s definition (1993: 141, footnote 2): ‘δή
wies auf dasjenige was von dem Sprechenden als dem Hörenden gegenwärtig Vorliegendes ....
vorgestellt wird hin’. At M en. 87e δή ‘no doubt contains sarcasm’, as Bluck ad loc. observes.
17 Dodds explains γε as: ‘“yes and ...”, adding an unexpected item to the enumeration’, but this
seems rather ad hoc.
18 In neither case could αΰ be replaced by δέ: apart from semantic considerations καί δέ does not
occur in Attic, and καί ... δέ never contrasts sets, but only single items.
 ALBERT RUKSBARON

I have discussed this first group of examples together because they share one crucial
syntactic characteristic: of the two particles present δέ can be left out, just as δή and
γε in the combinations καί ... δή and καί ...γε.19 In other words, again just as in the
latter combinations, it is καί which is the connector.

With respect to my second group the situation is altogether different, since both καί
and δέ may, in principle at least, each on their own operate as connectors at clause
and sentence level. A further, pragmatic, problem is that, whether δέ is taken adver­
bially or as a connector, in both cases it would mark a Topic shift; cf. fn. 11 and see
further below. Yet the syntactic similarity with καί ... δέ of the first group strongly
suggests that καί is the connector. Also, in many cases the context simply excludes
die meaning !and/but also’. And let me add that on the basis of the results achieved
so far I considered it a matter of methodological soundness to try and analyse these
cases of καί ... δέ in the same way.

3. κ α ί ... δέ connects clauses or sentences

3.1. Preceded by τ ε or κ αί

My first examples contain the correlative construction τε ... καί ... δέ. As in the case
of the combination καί ... καί ... δέ discussed above, the presence of τε indicates
that καί should be taken as the connector. Consider (22) and (23):

(22) καί τον πόλεμον αύτόθεν ποιεΐσθαι οΰπω έδόκει δυνατόν είναι, πριν αν
ιππέας τ ε μεταπέμψωσιν έκ των 'Αθηνών καί έκ των αύτόθεν ξυμμάχων
άγέιρωσιν ..., κ α ί χρήματα δ έ αμα αύτόθεν τε ξυλλέξωνται καί παρ'
’Αθηναίων έλθη, των τ ε πόλεων τινας προσαγάγωνται ..., τά τ ε αλλα ...
παρασκευάσωνται ...
(‘... and it seemed as yet impossible to carry on the war from this base until
they should send to Athens for horsemen, besides collecting them from their
allies in Sicily .... And they wanted at the same time to collect money from the
island itself and to have a supply come from Athens; also to bring over some
of the cities ...; and to prepare other things’, Th. 6 . 7 1 .2 )

Classen-Steup remark: ‘ιππέας τ ε steht erst zu κ α ΐ χρήματα δ έ in Beziehung;

wegen der entfernteren Verbindung wird das καί zu καί ... δέ verstärkt ...’20
19 In (21) αΰ cannot be left out as easily, probably because, as I argued above, the items are seman­
tically opposed. Without αΰ there would be a suggestion that ψυχή belongs to the same category as
σκεΰος and σώμα. This effect is even clearer in φιλόλογος ... κα ί αΰ μισόλογος, or χρήσιμόν τε
κα ί ωφέλιμον κα ί άχρηστον αΰ καί βλαβερόν, quoted above. Unlike δέ (cf. fn. 12), αΰ probably
operates at the representational level,
^ T w o other examples o f τε ... καί ... δέ are: ό δέ Κυαξάρης ... αύτός τε έμεθΰσκετο ..., καί
τούς άλλους δε Μηδους ώετο παρείναι έν τ<£ στρατοπεδφ ... (‘Cyaxares ... got drunk; and he
supposed that the rest o f the Medes were all in camp ...’, X. Cyr. 4.5.8), and: καί ο ί τε άλλοι προ-
 ADVERB OR CONNECTOR? 197

This is a particularly revealing example, because of the balanced structure of the

sentence, a balance brought about by the subdivision of the temporal πρίν-clause into
four members, marked by τε ..., καί ... δέ ά μ α,... πε and again τε (printed in bold
type). Observe that Classen-Steup’s analysis of this example is quite different from
what they suggest in their general note on καί ... δέ at 1.132.4, also quoted above:
Tn der Verbindung der Partikeln καί ... δέ ... ist καί nicht Kopula, sondern a u c h . ’
The change of opinion can be easily understood, for the coordination with the three
re ’s makes it virtually impossible n o t to consider καί here the connector.21 Their
view that δέ has been added to ‘strengthen’ καί because of its distance from τε is
entirely a d h o c . Rather, δέ should be given the ‘on the other hand’ or ‘for that
matter’ meaning discussed above; it explictly sets off χρήματα from ιππέας, and
marks, thus, a Topic shift.22
The same effect is present in (23), an example of καί ... καί ... δέ:

(23) καί ό πατήρ άεί λέγει καί συ φής κ αί οί άλλοι δέ πάντες όμολογοΰσιν,
ώς ...
(‘And my father always says, and so do you, and all the rest agree, th at...’, X.
Cyr. 3.3.19)

Note also that καί cannot be omitted, while δέ can.

3.2. Other cases

The example from Sophocles (without preceding τε) that played such an important
role in Denniston’s discussion of καί ... δέ, should be analysed in the same way:

βύμως τφ Τελευτίςι ύπηρέτουν ... καί ή τ<£ν Θηβαίων δέ π όλις ... προΘύμως συνέπεμψε καί
ύπλίτας καί Ιππέας (‘And all the states gave their hearty support to Teleutias... while the Theban
slate in particular... eagerly sent with him both hoplites and horsemen’, X. HG 5.2.37). In Pi. Cri.
48b: οΰτός τ ε ό λόγος ... έμοιγε δοκεί έ τ ι όμοιος είν α ι κα ί πρότερον καί τόνδε δέ (om. βδ)
αΰ σκόπει, there is a ‘slightanacoluthon’, as Burnet ad loc. observes, ‘since the κ α ί (b4) intro­
duces a changed construction’. Bumet is silent on δέ. Note the presence of αΰ reinforcing δέ, also
found in e.g. κα ί άλλο δέ τ ι αΰ, X. C yr. 4.5.46.
21 It might be argued that, in (22), (χρήματα) δέ corresponds with (ιππέας) τε. However, this is
unlikely in view of the position of δέ. For δέ to correspond with τε one would rather expect the
order άμα δέ κα ί, as in: κατά τε τό δίκαιον ΰπεδέξαντο την τιμωρίαν...... άμα δέ κα ί μισεί
τών Κερκυραίων (‘The Corinthians undertook the task, partly on the ground of right..., partly also
through hatred of theCorcyraeans'),Th. 1.25.3. Ο π τέ... δέ see further K-G 2,244 Anm. 3, Dennis-
tonp. 513.
22 Besides καί ... δέ άμα Thucydides also uses άμα δέ κα ί to connect clauses and sentences, as
in: κατεσκευάσαντο δέ ... πολλοί .... άμα δέ και τών προς τον πόλεμον ήπτοντο (‘Many also
established them selves.... And while all this was going on, (the Athenians) applied themselves to
the war Th. 2.17.4). Here, δέ must be taken as the connector, since κ α ί in this position cannot
possibly function as such; naturally, therefore, this combination is not found between single words,
as in ex. (1). The difference with καί ... δέ άμα at clause level, as in (22), probably is that in
2.17.4 καί has the whole clause that follows in its scope, while in (22) the emphasis is primarily
on χρήματα. Similar differences ace found with καί ... δέ ώσαύτως, sec the Appendix.
 ALBERT RIJKSBARON

(24) οίς γάρ ή γνώμη κακών

μήτηρ γένηται, πάντα παιδεύει κακούς,
καί σου δ' έγωγε θαυμάσας έχω τάδε,
χρήν γάρ σ’ κτλ.
(‘For when the mind hath once become a parent of evil, it teaches men
to be evil henceforth. And in thee, too, this conduct moves my wonder’,
S. P h . 1360ff., Philoctetes addressing Neoptolemus; transi. Jebb)

We have seen above that Jebb takes καί here as ‘also’, and δέ as ‘and’; he adds that
‘there is an emphasis on the intervening word.’ But this interpretation does not suit
the passage, for it entails that Philoctetes has voiced amazement in the preceding lines,
which is not the case. In those lines he is tormented by the thought of having to deal
with his enemies, the Atreidae and Odysseus. I think that καί ... δέ signals that
Philoctetes now turns to Neoptolemus himself: ‘And as for you, I am amazed at your
present behaviour’, viz. of trying to persuade him to go to Troy. The ‘emphasis on
the intervening word’ is not due, then, to καί but to δέ. As in examples (22) and (23),
δέ is here a marker of Topic shift.23
The same effect of καί ... δέ is present in:

(25) ευ ουν 'ίσθι άτι έγώ ταΰτα άκούων χαίρω οτι εύδοκιμέϊς, καί συ δέ ήγοΰ
με έν τοΐς εύνούστατόν σοι είναι
(‘So let me tell you that I rejoice to hear this and to know you have such a
good reputation; and you in return must count me as one of your warmest
well-wishers’, PI. L a . 181b; Lysimachus addressing Socrates)

Again, καί σύ δέ cannot possibly mean ‘and you too (must count me as one of your
warmest well-wishers)’, because there has been no mention, in the preceding clause,
of someone else who considered the speaker (Lysimachus) well-wishing. Rather, the
construction is as follows. By its position έγώ approaches έγώ μέν, and we expect a
second verb governed by on. But then the construction changes, and there starts an
independent clause, introduced by καί, which connects ήγοΰ with ίσθι. At the same
time δέ causes σύ to contrast with έγώ in the ou-clause. Lamb’s ‘in return’ captures
the effect pretty well.

23 A similar Topic shift marked by καί ... δέ occurs e.g. a tE . El. 1117: El. τ ί δ’ αΰ πόσιν σόν
άγριον ε ις ήμάς έχεις; / ΚΙ. τρόποι τοιοΰτοί' κα ί σύ δ' αυθάδης έφυς (‘And me—why let your
husband treat me so fiercely? Kl. That is his way; you loo are seif-willed, you know’, iransl.
Cropp). Dcnniston ad loc. notes: ‘καί ... δέ is rare in tragedy. Here, and in some other places, the
combination is only an apparent one, δέ being the connective and κα ί meaning “also”: “and you
too” (Particles, p. 200).’ But this is impossible, for 'too' only would make sense if άγριος and
αύθάδης were synonymous. For the same reason Cropp’s translation should be rejected: Aegisthus
is not αύθάδης. W e should rather interpret: 'and you, in turn,...’ Cf. also exx. (25) and (28) below.
- In Theophrastus’ Characters καί ... δέ is the means par excellence to achieve a Topic shift, and
thus to structure the description of a given type, cf. e.g. 23 'Αλαζονεία (3: κα ί συνοδοιπόρου δέ,
4: κα ί γράμματα δέ, 6: κα ί άγνώτων δέ, 7: κα ί προσελθών δ'). Note that Diels, in the Index
verborum to his edition, takes καί in κ α ί... δέ as = et, not as = edam.
 ADVERB OR CONNECTOR? 199

The non-connective status of δέ may seem less apparent in a case tike (26):

(26) ΣΩ .Ό ταν άρα δυοιν καλοίν θάτερον κάλλιον η, ή τφ έτέρω τούτοιν ή

άμφοτέροις ύπερβάλλον κάλλιόν έστιν, ή τοι ηδονή ή ώφελίςι ή
άμφοτέροις.
ΠΩΛ. Πάνυ γε.
ΣΩ. Κ αί όταν δέ δή δυοιν αισχροίν τό έτερον αϊσχιον ή, ήτοι λύπη ή
κακφ ύπερβάλλον αϊσχιον εσταν
(‘Thus when of two fair things one is fairer, the cause is that it surpasses in
either one or both of these effects, either in pleasure, or in benefit, or in both. -
Certainly. - And again, when one of two foul things is fouler, this will be due
to an excess either of pain or of evil’, PI. G r g . 475a)

‘And again’ is Lamb’s translation, which suggests that he takes καί as the connec­
tor. I think, in fact, that καί expresses the idea that Socrates continues the same line
of reasoning, while δέ signals that this procedure is now applied to a different item;
the δταν-clause serves to introduce the new Topic.24 One might be tempted, perhaps,
to take δέ as the connector, with καί as = ‘also’: ‘and/but also when’. Thereby,
however, δέ would be given the position of a sentence connector, although it occurs
in a subordinate clause. There are, to be sure, some instances where γάρ is used in
this way (e.g. κ ε ί μή γάρ εστιν ό θεός οΰτος, ... / παρά σοί λεγέσθω (Ε. Β α .
333)), but this construction is exceptional, and probably confined to fixed combina­
tions like καί ε ί (μή); cf, E. P h o . 946, Ar. E c c l. 888.25 Be that as it may, there are no
examples of καί Subordinator δέ where δέ m u s t be taken as the connector; see Hdt.
4.68.4 (καί ήν δέ (καί om. d, Hude), corresponding with καί ήν μέν), X. C y r .
8.2.25 (καί οπότε δέ; actually καί = ‘also’ excluded), PI. R e s p . 436e (καί όταν δέ
(καί Galenus, om. AFDM)), [PL] B r y x . 400e (καί δσα δέ; καί = ‘also’ excluded).
The status of δέ in e.g. καί δσα δέ is not different from that of αύ in καί δσα αΰ,
e.g. Th. 1.91.5. A particularly clear example of the adverbial use of δέ in a dependent
construction is:

(27) ταΰτα γάρ ούχ ώσπερ ε ί ..., καί ε ί νόσου δέ πολεμίοις έμπεσούσης
κρατήσειεν, ευτυχέστερος μέν αν ε’ίη, στρατηγικότερος δέ οΰδέν α ν
24 Observe that while καί connects the sentence as a whole with the preceding sentence, δέ oper­
ates at the level o f the temporal clause only, contrasting this clause with the previous öxav-clause.
Bakker (1993: 287ff.) believes that also in cases like ώς δέ ώρα... έωυτόν έμακάρισε (Hdt. 7.45),
δέ just belongs to the temporal clause. But this is impossible; in accordance with Wackemagel’s
Law δέ appears in second position, but it is, o f course, still a sentence connector. To pul it other­
wise: in (26) δέ appears in the second position of the clause, in ώς δέ ώρα... έωυτόν έμακάρισε
in the second position of the sentence.
2-* More common arc κ α ί γάρ ε ί (e.g. thrice in Plato, 5x in Isocrates, 20x in Demosthenes), ε ί γάρ
κα ί (e.g. once in Herodotus, 4x in Plato, 3x in Demosthenes). I have found no examples of καί
οταν/έπειδάν (öxe/έπειδή) γάρ. Instead, one finds καί γάρ öxav (PI. R. 343e), etc. I should per­
haps add that it is far from certain whether κα ί in καί γάρ εί, κ α ί γάρ όταν etc. always means
‘also’; καί (...) γάρ is probably the most Protean of all particle combinations.
 ALBERT RUKSBARON

(‘These are the marks that distinguish him (: Agesilaus) from ... or from the
man who wins victory through an outbreak of sickness among the enemy, and
adds to his success but not to his knowlegde of strategy’, X. A g e s . 10.1)

Here δέ exceptionally modifies a genitive absolute w ith in a καί ε ί-clause. Of course,

5é cannot operate as a connector here, since this would amount to ignoring two syn­
tactic boundaries, viz. those effected by ε ί and by the genitive absolute.

The position between καί and δέ may also be taken by a p a r t i c i p i u m c o n iu n c tu m ,

cf.:

(28) καί θέλων δέ τούτων πέρι σαφές τι είδέναι..., έπλευσα καί ές Τύρον ...
(‘Moreover, wishing to get clear knowledge of this matter ..., I took ship to
Tyre...’, I-Idt. 2.44.1)
Abicht ad loc.: ‘Auf dem zwischen καί und δέ eingeschobenen Worte liegt...
der Ton.’ Stein ad loc.: ‘Nicht verschieden von θέλων δέ καί. Das im
Attische geläufige καί—δέ zur Betonung des Parentheton findet sich bei H.
noch nicht.’

This passage is preceded by two sentences introduced by so-called ‘progressive’ καί

μέν, and followed by a sentence introduced by καί ειδον. This definitely suggests
that in our example κ α ί is the connector. From Abicht’s remark it docs not appear
whether he takes καί as = ‘also’. Be that as it may, this meaning is excluded for the
same reasons as those mentioned above in the discussion of examples (24) and (25).
To arrive at ‘also’, one would expect Herodotus to have spoken about willing other
things in the preceding context, but this is not the case.26 Here, too, καί ... δέ marks
a shift, in this case of the ‘setting’ or ‘attendant circumstances’: after the speculative
reasoning of the two καί-clauses in §§ 2-3, καί ... δέ marks the transition to a more
practical step in Herodotus’ argument concerning the name ‘Herakles’. As for Stein,
I wonder why Herodotus did not simply use δέ καί, if it all amounted to the same.27
A more difficult example is:

26 It might be argued that in (26) (and elsewhere) the focus of καί is not the word that follows but
the clause as a whole: ‘and also’ = ‘and further’. But such an approach to κα ί ... δέ runs counter to
the syntactic considerations mentioned above, which indicate that it is κα ί that is the connector.
Nor arc there parallels for such a use of καί in other combinations. To be sure, καί in καί γάρ may
be used in this way (e.g. X. An. 1.1.8), but this does not occur with καί ... γάρ separated, where in
general the focus is on the word following καί. For all that, ‘and further’ for κ α ί... δέ is often not
inappropriate, but in such cases the ‘further’ nuance is conveyed by δέ, not by καί. Notice, in this
connection, that Dcnniston renders both καί γάρ (p. 109, in some uses) and καί τοίνυν (p. 565)
by ‘and further’, in both cases with καί as the connector. Incidentally, part of the confusion created
by Denniston’s rendering of καί ... δέ by ‘and also’, both when he takes κα ί and when he takes δέ
as the connector (see section 1), may be due to his not distinguishing the different ‘also’s’ of
English, for which cf. Quirk et al. (1985: 604ff.).
27 Observe that in Ξέρξης δέ ... δρησμόν έβοΰλευε· θέλων δέ μή έπίδηλος ε ίν α ι ... χδμ α
έπειράτο διαχοΰν ... (Hdt. 8.97.1) the δέ-sentence contrasts not just pragmatically, but semanti­
cally with the preceding sentence. Here, κ α ί... δέ would seem excluded.
 ADVERB OR CONNECTOR? 201

(29) Ίώ Πρασιαί ... / ώς άπολεϊσθε τημερον.

(O ho Prasiae... how you’ll be murdered today! ’)

Ίώ Μέγαρα Μέγαρ’, ώς έπιτρίψεσθ' αύτίκα

(Ή ο Megara, Megara, how you will be crushed presently ...’)

Ίώ Σικελία, καί συ δ' ώς άπόλλυσαι

(O ho Sicily, what a doom is yours too!’, Ar, Pax 243-4, 246, 250; transl.
Sommerstein)28

In view of the preceding lines, it is tempting, of course, to take this καί adverbially
and δέ as the connector. However, on the analogy of καί σου δέ in (24) and καί συ
δέ in (25) (also E. E l . 1117 (note 23)), as well as καί ύμΐν δέ, καί ύμας δέ in (37)
and (38) below, I think that, once again, καί should be taken as the connector, just as
καί is the connector in καί ... γε in similar greeting and farewell formulas, e.g.:

(30) Καί χαΐρ ε πόλλ’, ώδελφέ. ΔΙ. Νή Δία καί σύ γε

ύγίαινε.
(‘And fare thee well, my brother. Di. And to you good cheer’, Ar. R a . 164;
transl. Rogers)

(31) ό δέ Κροΐσος ώς είδε τον Κΰρον, Χαιρε, ώ δέσποτα, έφη. (-) Καί σύ γε,
έφη, ώ Κροΐσε
(‘And when Croesus saw Cyrus, he said: “I salute you, my sovereign lord (-)
And I you, Croesus’” , X. C y r . 7.2.10)

Cf. also Ar. L y s . 6, P a x 718, E c c l. A l l . It should also be observed that in (29), as in

many other cases as we have seen, καί cannot be omitted, while δέ can. This again
strongly points to δέ not being a sentence connector, but an adverb: ‘And you, on the
other hand/for that matter’.29

3.3. κ α ί ... δ έ = ‘and also/too’?

With regard to example (25) I observed that ‘and you too’ for καί σύ δέ is
excluded, given the lack of someone else who considered Lysimachus a well-wisher;
a similar reasoning applies to e.g. καί σου δ' in (24) and to examples (37) and (38)
below. In more general terms one might say that for καί = ‘also’ to be acceptable the
28 Van Leeuwen on Pax 250 only comments that καί ... δέ, which, according to him, often has
dubious ms authority in Aristophanes, ‘hoc certe loco... suspicione vacat’. Platnauer has no remark
at all.
29 In view of the syntactic similarity I am inclined to follow the same reasoning at Pax 523 (ώ
%tììp’ 'Οπώρα, κα ί σύ δ' & Θεωρία). Another possibility would be to lake both καί and δέ as
adverbs, e.g. as in Engl.: ‘you too, for that matter’. But this would leave die clause/sentence with no
connecting panicle at all.
 ALBERT RIJKSBARON

states of affairs of the two clauses or sentences must be identical or similar, while the
entities involved, are different. Indeed, this explains why in example (29) ‘and you
too’ is in principle fully acceptable: ώς άπόλλυσαι repeates, so to speak, ώς
έπιτρίψεσθ' of line 246, and κ α ι ... δέ expresses the idea that the state of affairs
mentioned in the preceding sentence applies to a second item as well.30 There are
many comparable passages where our first inclination will be, I think, to take καί as
= ‘also’. Consider:

(32) άπιόντα δέ τόν Κΰρον προύπεμπον άπαντες ... καί ούδενα έφασαν οντιν’
ού δακρύοντ’ άποστρέφεσθαι. καί Κΰρον δέ αυτόν λέγεται συν πολλοΐς
δακρύοις άποχωρήσαι
(‘And everybody ... escorted him ..., and they say that there was no one who
turned back without tears. And Cyrus also, it is said, departed very tearfully’,
X. C y r . 1.4.26)

(33) Δοκεΐ γοΰν, εφη ό Κρντόβουλος, οικονόμου άγαθοΰ είναι εΰ οίκεΐν τον
έαυτοΰ οίκον. Ή καί τόν άλλου δέ οίκον, εφη ό Σωκράτης, ε ί έπιτρέποι
τις αΰτω, ούκ αν δύναιτο, εί βοΰλοιτο, ευ οίκέίν, ώσπερ καί τόν έαυτοΰ;
(“‘Well, I suppose that the business of a good estate manager is to manage
his own estate well.” “Yes, and in case he were put in charge of another
man’s estate, could he not, if he chose, manage it as well as he manages his
own?”’, X. O e c . 1.3)

(34) καί σπείσας καί εϋξάμενος επιε, καί οί άλλοι δέ οί περί αυτόν ούτως
έποίουν
(‘And when he had poured a libation and prayed, he drank; and the rest, his
staff-officers, followed his example’, X. C y r . 7.1.1)

Observe that in (33) the predicate and the object of the second clause (τόν οίκον εΰ
οίκεΐν) literally repeat those of the first clause (εΰ οίκεΐν τόν οίκον), while in (32)
there is lexical variation (συν πολλοΐς δακρύοις άποχωρήσαι as against οΰδένα
30 Note that the wording of the καί ... δέ-sentence is not an exact duplicate of the preceding sen­
tence. In fact, this is very rare, mostly there is some degree of variation, cf. c.g. (32) and (34) below.
There are several instances of this type in Homer, e.g. οΰτος γάρ δή όνήσει έϋκνήμιδας ’Αχαιούς,
/ καί δ' αύτός δν θυμόν όνήσεται, αί κε φύγησι (‘He shall be the one to gladden the strong-
greaved Achaians, and to be glad within his own heart, if he can come off, II. 7.173; transi. Latti-
more). Some other examples are//. 12.272,13.73, O d. 6.60, and, without αύτός or άλλος,//. 7.113,
375, O d. 13.302. Whenever Ameis-Hentze comment on such cases they always interpret καί as =
‘also’. Kiihner-Gerth, on the other hand, always take δέ as adverbial and καί as the connector. I
think K-G arc right. Thus, in //. 7.173 καί 5' means ‘and apart from that, and moreover’; ‘and, for
that matter’ is also possible. The point is that, just as in Attic Greek, there arc many cases where
καί δέ cannot p o ssib ly be taken as ‘and also’, with the emphasis on the following word, e.g. II.
17.351, 19.178,20.28, O d. 4.391,13.302. In the last three cases A-H were aware of the problem, for
there they take καί at sentence level, e.g. II. 20.28 καί δέ τέ μιν καί πρόσθεν ‘das erste καί
(bezieht) sich auf dem ganzen Satz’ (similarly Edwards ad loc.), but this is highly unlikely, cf.
footnote 26. It should be noted, finally, that καί in the combination καίτε is taken a/so as= ‘also’
by A-H, whereas in all probability it is the connector, τε being adverbial; cf. Ruijgh (1971:764ff.).
 ADVERB OR CONNECTOR? 203

övuv' ού δακρύοντ άποστρέφεσθαι). In (34), finally, the second clause has the
pro-forms ούτως έποίουν, representing the three states of affairs of the preceding
clause. Observe also that the examples contain a form of αυτός ‘self’ or άλλος
‘other’. In fact, many passages where an ‘and also’ interpretation seems to impose
itself contain such forms, not surprisingly perhaps, since ‘self’ and ‘other’ are espe­
cially at home in clauses or sentences that are semantically (near-)repetitions o f the
preceding sentence or clause.31
But is this ‘and also’ interpretation inevitable? I think not. Assigning an overall
‘and on the other hand’-interpretation to κ α ι ... δέ is quite possible, perhaps most
clearly in (33), where κ α ι... δέ might be rendered more specifically by ‘and again’,
while both in (32) and in (34) ‘and in turn’ is more appropriate. As in the other
cases, then, δέ marks a Topic shift. I prefer this analysis to distinguishing a separate
use of καί ... δέ where καί - ‘also’.32
Another obvious example of ‘and also’ would seem to be:

(35) έπυνθάνοντο δέ καί ές τους Είλωτας πράσσειν τι αυτόν, καί ήν δέ

ούτως- έλευθέρωσίν τε γάρ ΰπισχνεΐτο αύτοΐς .... άλλ' ούδ’ ώς ...
ήξίωσαν ....
(‘They were informed also that he was intriguing with the Helots; and it was
even so, for he was promising them freedom ... But not even then ... did they
think it best...’,Th. 1.132.4)

where ούτως takes up πράσσειν τι, Indeed, as in the other cases of καί ... δέ in
Thucydides, Classen-Steup and others take καί as = ‘also’. We have already seen,
however, that this is very unlikely at 2.36.1 (ex. (3)) and 6.71.2 (ex. (22)), and here,
too, there is no need to take it in this way. In the preceding context we have been told
how the Spartans distrusted Pausanias because of his lifestyle, and discovered all
kinds of incriminating evidence. The έπυνθάνοντο-clause mentions another poten­
tially damaging fact Now καί ήν δέ ούτως plus lite following sentence give Thucy­
dides’ comments on this clause, by way of a parenthesis; δέ emphasizes ήν, and
marks the shift from ‘rumours’ to ‘reality’: ‘and, as a matter of fact, this was the
case; fo r...’ On this analysis, the έπυνθάνοντο-clause is answered by άλλ' ούδ' ώς
etc.33

31 That the presence o f αύτός alone is not sufficient for this effect appears e.g. from X. Cyr. 2.4.17.
32 In itself such a double use is not impossible, cf. καί (...) γάρ, where καί may or may not be
‘also’.
33 To bring this out the whole o f καί ήν δε ούτως ...ξυγκατεργάσωνται should, in fact, be put
between parentheses. - The remaining examples from Thucydides can be analysed in a similar way,
4.24.2: ‘and they themselves, for that matter’; 7.56.3: ‘and, as a matter o f f a c t , D u . ‘En trouwens,
de strijd was memorabel’; 8.67.3: 'and, on the other hand’, ‘and, in turn’.
 ALBERT RUKSBARON

3. 4. ο ΰ δ έ ... δέ

The next example to be discussed contains ουδέ ... δέ. Recall that according to
Denniston (p. 203) ‘... δέ is clearly the connective, and οΰδέ is adverbial.’

(36) καν ούδέν μεντοι ουδέ τούτον παθέΐν έφασαν, ούδ' άλλος δέ των
Ελλήνων ... έπαθεν οΰδείς ούδέν
(‘yet it was said that even he was not hurt in the least, nor, for that matter, did
any other single man among the Greeks get any hurt whatever ...’, X. A n .
1.8.20; ‘for that matter’ is present in the Loeb- transl.)

I do not think Denniston is right. Instead of continuing, after ούδέ τούτον παθεΐν
έφασαν, with ούδ’ άλλον (δέ), Xenophon starts a new sentence, which no longer
represents other people’s words, but voices his own comments. By the transition to a
new main verb there is a slight anacoluthon, but ούδ’ (άλλος) still corresponds with
οΰδέ (τούτον). In other words, οΰδέ is the connector and δέ should be taken adver­
bially. And it should be noted that, just as in the case of κ α ι... δέ, δέ can be omitted,
while ούδέ must be present34

3.5. μεν ... κ α ι ... δέ

To conclude I present two cases where καί ... δέ combines with μέν, a combination
also mentioned by Denniston (p. 203), who does not tell us, however, whether he
takes καί or δέ as the connector.

(37) έγώ μέν τοίνυν εύχομαι πρίν ταΰτα έπιδειν ύφ’ ύμών γενόμενα μυρίας
έμέ γ ε κατά της γης όργυίας γενέσθαι. καί ύμίν δέ συμβουλεύω ...
πειρασθαι των δικαίων τυγχάνειν
( ‘For my part, therefore, I pray that sooner than live to behold this deed
wrought by you, I may be laid ten thousand fathoms under the ground. And to
you my advice is th a t... you endeavour to obtain your just rights ...’, X. A n .
7.1.30)

(38) Έγώ μέν ούν ... κατέστησα τε τουτουσί εις αγώνα κ αί ύπό την ύμετέραν
ψήφον ήγαγον, καί ύμάς δέ χρή ... ψηφίσασθαι τά δίκαια
(‘I therefore ... have brought them to trial and submitted them to be judged by
you. It is now your duty to render ihe verdict which justice demands ...’, [D.]
59.126)

34 This passage is mentioned explicitly by K-G, 2 ,2 6 1 , as an instance of adverbial δέ. - Strangely

enough, Denniston gives Anstot. ΕΝ 1120a31 as another instance of adverbial ούδέ, although ihe
presence in the preceding clauses of ού and ούδέ definitely points to ούδέ being the connector.
 ADVERB OR CONNECTOR? 205

How should these passages be analysed?35 (Observe that the translations suggest
that καί is not taken adverbially, rightly, as we will see). First, there can be little doubt
that δέ is meant to contrast ύμΐν, and ΰμας, with έγώ μέν. Next, it would seem that
καί can be easily omitted, which would point to δέ being the connector. And a
sequence like έγώ μέν τοίνυν ε ύ χο μ α ι... ύμΐν δέ συμβουλεύω is, of course, per­
fectly acceptable. Syntactically, then, δέ might be the connector. But for semantic rea­
sons καί must be taken as such. The point is that if καί is taken adverbially, i.e. as =
‘also’, with the emphasis on the pronoun, there is a semantic clash between the two
sentences, a phenomenon we have already met in cases like (24) and (25). For to
interpret (37) as ‘and I advise you, too, to try and get...’, the first sentence must nec­
essarily have mentioned other persons who got this advice. But this is not the case.
Moreover, to express, in the appropriate context, the Greek equivalent of the sentence
just mentioned, the order should rather be συμβουλεύω δέ καί ύμΐν. The same anal­
ysis applies to (38): ‘you, too, must vote is, in the context, nonsense. In both
cases δέ has its by now familiar ‘on the other hand, in turn’ value.

4. Conclusion

In the introduction I noted that Denniston to a large extent, and Kuhner-Gerth wholly,
analyse the particle combination καί ... δέ as consisting of connective καί and
adverbial δέ, roughly meaning: ‘and on the other hand’. Neither Kühner-Gerth nor
Denniston, however, provided any arguments for their view. Many commentators, on
the other hand, if they comment at all, take καί as =‘also’ and δέ as the connective.
The upshot of the present paper is that Kühner-Gerth and Denniston were right. Its
main points are as follows.
The combination καί ... δέ occurs in two different syntactic surroundings.
Firstly, it may connect single nouns, adjectives and verbs, type: δίκαιον καί πρέπον
δέ. In these phrases καί cannot be omitted. This fact, as well as the existence of cor­
relative καί ... κ α ί ... δέ in phrases like καί ούτος καί άλλοι δέ ποιηταί and καί
τοξότας καί σφενοονήτας καί ιππέας δέ, shows that καί, rather than δέ, should
be taken as the connector. The function of δέ is a pragmatic one: it presents the item
concerned as distinct from the preceding item(s), and indicates that the καί ... δέ item
should be considered in its own right.
Secondly, καί ... δέ is used to connect clauses and sentences. Here, in principle
both καί and δέ are suitable connectors. Again, however, the existence of correlative
constructions, e.g. πριν αν ιππέας re μεταπέμψωσιν..., καί χρήματα δέ άμα ...
ξυλλέξωνται..., and κ αί ό πατήρ ά εί λέγει καί σύ φής καί οί άλλοι δέ πάντες
όμολογοΰσιν indicates that it is καί that should be taken as the connector. The gen­
eral value is the same as that mentioned above: ‘and on the other hand’. Often, κ α ί...
δέ marks a Topic shift.
35 Another example of μεν ... καί... δέ is PI. P rot. 331b.
 ALBERT RUKSBARON

I further argued that those cases where at first sight an ‘and also’ interpretation
for κ α ι ... δ έ seems plausible can very well be analysed in conformity with the ‘and
on the other hand’ approach. Since there are no cases where καί m u s t be taken as
‘also’, while there are many cases where καί c a n n o t p o s s i b l y be taken in this way, I
conclude that, in the combination καί ... δ έ , καί is the connector and δ έ an adverb.

Appendix: κ α ί ... δ έ αχταύτως, and other constructions involving ώ σ αύτω ς

In Plato, we find a number o f combinations involving ώσαύτως that p rim a fa c ie are rather similar:
κ α ι ... ώσαύτως, κ α ι... δέ ώσαύτως, καί αΰ ... ώσαύτως and ωσαύτως δέ καί, as in the follow­
ing examples:

(i) δοκεΐ οΰν σοι .... άπολείπειν άν τ ι ό τοιοΰτος αρετής, εΐπερ ε ίδ ε ίη τά τε άγαθά
πάντα καί παντάπασιν ώς γ ίγνεται καί γενήσεται κα ί γέγονε, κ α ί τά κακά ώσαύτως
(’Now do you th in k ... there could be anything wanting to the virtue o f a man who knew all
good things, and all about there production in the present, the future, and the past, and all
about evil things likewise?’, PI. L a . 199d)

(it) Ε ίσίν δή κατά τούς vßv λόγους ψ ευ δείς έν ταίς τών ανθρώπων ψ υ χαΐς ήδοναί,
μεμιμημέναι μέντοι τάς αληθείς έ πί τά γελοιότερα, καί λύπαι δέ ώσαύτως
( ‘Hence we reach the result that false pleasures do exist in men’s souls, being really a rather
ridiculous imitation of true pleasures; and the same applies to pains, for that matter’, PI.
Phil. 40c6, transl. Hackforth; ‘for that matter’ added)

(iii) έν δέ ταύτη ούση τοιαύτη άνά λόγον τά φυόμενα φύεσθαι, δένδρα τε κα ί άνθη καί
τούς καρπούς· κ α ί αΰ τά δρη ώσαύτως καί τούς λίθους έχειν ... την τ ε λειότητα ...
('A nd in this fair earth the things that grow, the trees, and flowers and fruits, are corre­
spondingly beautiful; and so too the mountains and the stones are smoother ...’, PI. Phd.
11 Od)

(iv) ή τε οΰν Ιατρική ... πάσα διά τού θεού τούτου κυβερνάται, ώσαύτως δέ κα ί γυμνασ­
τική κα ί γεωργία·
(‘And so not merely is all medicine governed ... through the influence of this god, but like­
wise athletics and agriculture1, PI. Smp. I86e)

Actually, however, the differences between these four expressions are, I think, roughly as follows.
In examples (i) - (iii) κα ί functions as a connector. Example (i) is pragmatically unmarked;
κα ί corresponds with τε, and simply adds an item to which είπερ είδείη etc. applies as well, the
latter being summarized by the pro-form ώσαύτως. In example (ii) the speaker, by adding λύπαι by
means o f καί ... δέ, sets them o ff from the preceding item, and indicates that he is fully aware that
this item is relevant as well. Observe that in (ii) δέ can and κα ί cannot be omitted; conversely, in
the passage from Laches δέ might be added without affecting the syntax of the καί-clause. Example
(ii) is, then, pragmatically marked. In (iii), finally, the contrast between the two members is the
most marked, since it is semantic rather than merely pragmatic: while καί indicates that the line of
reasoning followed in the first member will be continued, αΰ emphatically signals that this reason­
ing will now be applied to a different set, that of non-growing items.36

36 As in the use of αΰ at word-level (see ex. (21)), αΰ contrasts sets that are semantic opposites.
O f course ώ εήδό ναι and λύπαι of (ii) arc also semantically opposed. But κ α ί ... δέ does not con­
trast them qua opposites. The pragmatic and semantic markedness of (iii) also appears from the fact
 ADVERB OR CONNECTOR? 207

As to example (iv), (his differs syntactically, of course, from the former three, since in this case
δε must be taken as the connector; for δέ corresponding with τε cf. fn, 20. Consequently, κα ί must
be taken adverbially. The difference with the examples where ωσαύτως occurs later in the clause may
be that in (iv) ώσαύτως, being in front-position, functions more emphatically as a pro-form: it
stresses that the speaker's main Topic still is διά τοΰ θ ε ο ί τούτου κυβερνάσθαι, a Topic intro­
duced at 186b:... ώς ... έπ ί παν ό θεός τείνει.

References

Bakker, EJ.
1993 ‘Boundaries, Topic and the Structure of Discourse: An Investigation of the
Ancient Greek Particle d é '. S t u d i e s i n L a n g u a g e 17, 275-311
Blomqvist, J.
1979 D a s s o g e n a n n t e κ α ί a d v e r s a t i v u m . Z u r S e m a n t i k e i n e r g r i e c h i s c h e n P a r ­
tik e l. Uppsala
Denniston, J.D.
19542 T h e G r e e k P a r t i c l e s . Oxford
Dik, H.J.M.
1995 W o r d O r d e r i n A n c i e n t G r e e k . A P r a g m a t i c A c c o u n t o f W o r d O r d e r
V a r i a t i o n i n H e r o d o tu s . Amsterdam: Gieben
Kroon, C.H.M.
1995 D i s c o u r s e P a r t i c l e s in L a t i n . A S t u d y o f nam, enim, autem, vero a n d at.
Amsterdam: Gieben
Kühner, R. - B. Gerth
1904 A u s f i i r h l i c h e G r a m m a t i k d e r g r i e c h i s c h e n S p r a c h e , Z w e i t e r T e il: S a t z l e h r e .
2 vols. Hannover and Leipzig
Pinkster, H.
1972 O n L a t i n A d v e r b s . Amsterdam
Quirk, R., S. Greenbaum, G. Leech, J. Svartvik
1985 A C o m p r e h e n s i v e G r a m m a r o f th e E n g l i s h L a n g u a g e . London: Longman
Rijksbaron, A.
1993 ‘Sur quelques différences entre ουτος ó (substantif), ουτος δέ ό (substantif)
et ό δέ (substantif) ουτος chez Hérodote’. L a l i e s 12,119-130
Ruijgh, C.J.
1971 A u t o u r d e τ ε é p iq u e . Amsterdam: Hakkert
1996 ‘La place des enclitiques dans l ’ordre des mots chez Homère d ’après la loi
de Wackemagel’. In: S c r i p t a m i n o r a I L Amsterdam: Gieben, 627-647
(orig. published in: H. Eichner & H. Rix (Hrsg.), S p r a c h w i s s e n s c h a f t u n d
P h i l o l o g i e : J a k o b W a c k e r n a g e l u n d d ie I n d o g e r m a n i s t i k h e u te . Wiesbaden;
Reichert, 213-233)
that in this example κ α ί is provided with a full-blown clause of its own, where ώσαύτως is not just
a pro-form, as in (i) and (n).
208 ALBERT RIJKSBARON

Sicking, C.MJ. & J.M. van Ophuijsen

1993 T w o S tu d ie s in A t t i c P a r tic le U s a g e . L y s i a s & P la to . Leiden: Brill
Slings, S.R.
1980 ‘KAI adversativum: Some Thoughts on the Semantics of Coordination’. In:
D.J. van Alkemade et al. (eds), L i n g u i s t i c S t u d i e s O f f e r e d to B e r t h e
S ie r ts e m a . Amsterdam, 101-125
 EMPHASIS AND AFFIRMATION
Some Aspects o f μην in Tragedy

Ger r y w a k k e r
Rijksuniversiteit Groningen

1. Introduction

It is a well known fact that the meaning and actual usages of the particles μήν, ή and
δή are a complicated matter. Terms such as emphatic, affirmative and confirmative are
often used more or less indiscriminately to characterise these particles. To take Den-
niston’s (1954: 329-58) description of μήν as an example, he argues (329) that ‘μήν
fulfils three functions: (1) as an emphatic particle: (2) as an adversative connecting
particle: (3) as a progressive connecting particle.’ As a first point of criticism it may
be said that it is nowhere explained how these three functions of μήν arc interrelated.
Denniston only observes that the first category, μήν as an emphatic particle, is ‘the
earliest sense’. In Homer this seems indeed the only (or nearly the only) use, the
connective use of μήν being non-existent (Denniston 329, 334; L f G r E s y . μάν/μήν).
But this observation does not explain how, at a synchronic level, these three widely
divergent, so-called functions or senses are interrelated. Nor is it clear what exactly is
meant by the term ‘emphatic’, or in which respect this emphatic μήν differs from
other particles, especially from γ ε and δή, whose main function is said to provide
emphasis as well. Note, in this connection, Denniston’s own remark (330): ‘It is dif­
ficult to grasp the exact difference in sense between μήν and the far commoner δή.’
Another, related, question is that of the difference between ή and μήν. Denniston
is not very outspoken on this point. In his introductory paragraph (279) to ή he
merely mentions the views of his predecessors Kiihner-Gerth (who consider the dif­
ference between ή and μήν or δή one of strength, see 2, 144) and Bäumlein (who
argues that ή expresses subjective certainty). The most recent treatment of this ques­
tion is that by Sicking.1Although Sicking has many excellent observations on the
frequent occurrence of both particles in certain collocations and on their preference
for conversational contexts, he, too, does not make clear the precise difference in
function and/or use between μήν and ή. As to μήν, Sicking notes (Sicking & Van
Ophuijsen 1993: 54): ‘The particle μήν seems to be at home in expressing the con­
trary of what the person addressed might either (1) suppose or (2) wish.’ As to ή,
Sicking accepts Ruijgh’s hypothesis that originally ή and ού were counterparts;

Sicking (1986: 132-3,137) and Sicking & Van Ophuijsen (1993: 51-7, 61-3).
210 GERRY WAKKER

accordingly, η is ‘used for affirming as opposed to denying (...) In communicating

with an audience just as the need to deny a proposition expressly may be felt when
there is a suspicion that the audience is disposed to assent to this proposition, so
there may be good reasons for explicitly affirming what the audience may be inclined
to disbelieve or to disregard, which is precisely the use of η.’ (Sicking & Van
Ophuijsen 1993: 55). I must confess that the exact difference between ‘expressing
the contrary of what the person addressed (= the audience?) might (= may be inclined
to?) suppose or wish’ (μην) and ‘affirming what the audience may be inclined to
disbelieve or disregard’ (ή) is not easy to see. Nor is it immediately clear in which
way these descriptions may account for the apparently different distributional fea­
tures of μην and ή.
In this article I will try to give an answer to the questions raised above. I will first
sketch a theoretical framework for the description of particles in general (section 2).
This framework will allow me next (section 3) to present a more coherent description
of μήν than the one offered by Denniston, and it will provide a basis to account for at
least some of the differences in meaning and use between μήν, ή, δή and γε in a sys­
tematic way. My corpus of examples is of necessity restricted. It consists of all
instances of μήν and ή, and of a representative number of instances of γε and δή, in
tragedy.2

2. The framework

Particles traditionally belong to that category of linguistic elements that are very diffi­
cult to describe: they usually fall outside the syntactic structure of the clause in which
they occur, and their meaning is elusive. In a strictly semantic approach it is difficult
to account for the diversity of - seemingly unconnected - uses of most particles. With
the relatively recent development of functionally oriented linguistic theories a new
(pragmatic) approach offered itself and has proven to be extremely rewarding in the
field of particle research.3 Within these theories language is considered primarily a
means of communication between a speaker and an addressee. It is the communica­
tive situation in which the utterance takes place that primarily determines its form,
rather than the ‘free choice’ of the speaker. This communicative or pragmatic side is

2 Since the use o f μήν and ή seems to differ per dialect and genre I have chosen only one genre, i.e.
tragedy, as my corpus. Moreover tragedy offers a great many examples of both particles. Exact
numbers arc difficult to provide given the many variae lectiones (μέν vs. μήν; ή vs. ή). Roughly
speaking, however, μήν occurs approximately 47 times in Aeschylus, 44 times in Sophocles, and
118 times in Euripides, whereas ή occurs about 65 times in Aeschylus, 95 times in Sophocles and
140 limes in Euripides.
3 For the necessity of a pragmatic approach o f particles see e.g. Levinson (1983: 100), Abraham
(1986: 87-100), Kroon (1992: 53-8; 1995: 34-57). As to Ancient Greek, pragmatic descriptions of
particles may be found in Hellwig (1974), Bakker (1986, 1988, 1993), Slings (1980), Sicking
(1986), Sicking & Van Ophuijsen (1993), Wakker (1994: 301-64; 1995). For some time to come
the monograph by Kroon (1995) will be the reference point for other particle studies in Latin as
well as Greek. She provides a refined analysis of the Latin particles nam, enim, autem, vero and at
within a completely worked-out discourse-pragmatic framework.
 EMPHASIS AND AFFIRMATION 211

also of prime importance for determining the ‘meaning’ of a particle. According to

recent research, roughly speaking two types of meaning should be distinguished:
r e f e r e n t i a l m e a n i n g (i.e. the meaning that contributes to the representation of an
event, a situation, an action etc.), and f u n c t i o n a l m e a n i n g which has to do with the
placing of the state of affairs described in the communicative (textual and non­
textual) context (Foolen 1991; Kroon 1992: 55-6; 1995: 41). Particles clearly
express the second type of meaning: they primarily have a pragmatic (rather than a
grammatical or strictly semantic, i.e. referential) function. From the point of view of
the addressee particles may, then, be considered a kind of road signs in the text which
help him keep track of the structure of the text or find out the communicative purpose
or expectations of the speaker. From the perspective of the speaker particles may be
described as a means of placing the unit they have in their scope into a wider per­
spective, which may be the interactional situation of which the text forms part or the
surrounding context (and its implications). As such, particles may, for instance,
enable the speaker to anticipate or respond to the (supposed) knowledge, presupposi­
tions, expectations or questions of the addressee, or they may mark the coherence
between different text units for the benefit of the addressee. If one studies particles
from this pragmatic perspective in a systematic way, one can considerably improve on
Denniston’s descriptions,4 for instance, because in this way one is able to make dis­
tinctions in the function and use of particles of which the primary function is said to
be the same, e.g. providing emphasis (γε, δή, μην, μέν).

Central to a pragmatic approach of panicles is the recent finding that every discourse
can be analysed at at least three levels.5

(1) discourse level type of relationship some

examples
representational semantic relations between ίνα, εί, έπεί,
the states of affairs in the άλλά, καί
represented world γε, περ
presentational functional relations between οΰν, δέ
discourse units
interactional the relation of a discourse αρα, που, δή,
unit to its non-verbal, τοι
communicative environment

4 Sometimes Dcnniston incorporates the reader/listener into his description (cf. e.g. p. xli/ii; lxxiv);
however, he docs not do so systematically.
5 For the difference between text ((lie product of writing or speech, detached, from its situational
context) and discourse (the dynamic process o f writing and speech within its situational context)
see Kroon (1995: 30n50). The framework presented here is based on Kroon (1995), Halliday
(1985), Schiffrin (1987).
212 GERRY WAKKER

Particles can be described and subcategorised in terms of the discourse levels at

which they function. In other words, the discourse function of particles is stated in
terms of the discourse level they are primarily concerned with.
The r e p r e s e n t a t i o n a l level of discourse is concerned with the representation of
some (real or imaginary) world. Particles which function at this level of discourse
signal relations among the entities or states of affairs that make up the represented
world. In Greek one may think of subordinating conjunctions like ε ί (‘if), 'iva (‘in
order that’); when used as coordinators Greek άλλά and καί, too, may be said to be
primarily representational particles. Also the so-called s c o p e - p a r t i c l e s belong to this
group. Scope particles may be defined as particles that mark the scope of (part of) the
utterance in the sense that they define its exact limits, i.e. they specify to whom/to
what/in which case etc. the utterance applies. This rough characterisation may be said
to hold for γε (limiting the applicability of the utterance to a t l e a s t the item modified
by γε, e.g. έκ γ’ έμοΰ ‘at least in as far as it depends on me’, S. A n t . 207), post-
homeric περ (limiting the applicability of the utterance to p r e c i s e l y a n d e x c lu s iv e ly
the item modified by περ cf. ε’ίπερ γυνή o ü‘<yes> exclusively in the case that you
are a woman’, S. A n t . 741) and for adverbial καί (extending the scope of the utter­
ance to a l s o the item modified by καί, e.g. ε ί καί δυνήση γ' ‘if you will also be able
to - i.e. besides having the intention to do if, S. A n t . 90).6
P r e s e n t a t i o n a l particles concern the question of how the represented world is
presented or staged by the language user. They signal, for instance, how one dis­
course unit is functionally related to another (in terms of central and less central
information units, explanatory or elaborating discourse units, etc.). Or they help the
addressee to keep track of where he/she is in the organisation of a monological
stretch of discourse in that they point to the functional links or boundaries between
the various information units. As such they serve to indicate digressions, resump­
tions, conclusions, discourse topic shifts, and so on. A Greek example is οΰν, which
marks a ‘new step’ in the argumentation or narration—a step, however, which is in
some way logically connected with the preceding context, for instance, as a new point
in the line of reasoning, or as a conclusion, or as a return to the main sequence of the
discourse after some side sequence such as a parenthesis or a comparison. Some­
times δέ, too, whose primary function is to mark a new text unit, signals a return to
the main sequence, especially in the combination νυν δέ (used after counterfactual
conditionals or comparisons to return to the real world/situation, cf. e.g. S. O T 258,
263). It goes without saying that often rather large stretches of discourse must be
taken into account in order to explain the presence of such particles.
The i n t e r a c t i o n a l level accounts for the fact that every coherent stretch of dis­
course is integrated into a specific interactional situation. Particles primarily function­
ing at this level may pertain to the involvement of the discourse participants (speaker -
addressee, narrator -reader) in the communicative situation, or to their commitment to
the message being exchanged; they may modify or specify the illocutionary intention,
or indicate the turn-taking system in a conversation. At this interactional level func-

6 S ee S icking (1986:13 4 -6 ); W ak k e r (1994: 307, 3 0 8 4 2 ) .

 EMPHASIS AND AFFIRMATION 213

tion the so-called m o d a l l a t t i t u d i n a l particles.7 These specify the attitude of the

speaker with regard to the (truth of the) proposition and the supposed knowledge,
expectations etc. of the addressee, cf. in Greek e.g. αρα (originally ‘surprisingly’:
the speaker signals his interest and/or surprise, in order to anticipate or elicit a possi­
ble reaction of surprise on the part of the addressee; then also ‘apparently’, to draw
attention to a sudden—and often surprised—awareness on the part of the speaker,
e.g. S. A n t . 1178, 1273), δή (Took, how interesting’; the speaker drawing the
addressee’s attention; the reasons for doing so may be manifold, but often it has to
do with indignation, cf. S. A n t . 726, 821), που (‘perhaps’; the speaker signalling his
doubt and thus agreeing beforehand with any possible doubt on the part of the
addressee), to i (T tell you’, T can assure you’; the speaker signalling that he con­
siders the proposition of special interest for the addressee, cf. S. A n t. 327-9).

3. The particle μην8 in tragedy

3.1. The so-called emphatic function of μην

Within the functional framework sketched above the so-called emphatic μην is best
described as an attitudinal particle that primarily functions at the interactional level of
discourse. In using μήν the speaker expresses his positive commitment to the truth of
the proposition; he indicates that he as it were personally guarantees its truth: ‘in
truth’, ‘really’. This insisting on the truth of the proposition is not a mere sign of
‘emphasis’, rather the speaker in this way anticipates a possible reaction of disbelief
on the part of the addressee,9 cf.

(2) Philoctetes: άλλ’ άντιάζω, μή με καταλίπης μόνον.

Neopt: θάρσει, μενοΰμεν. Phil, ή μενεις; Neoptol. σαφώς φρονεί.
Philoctetes: οΰ μήν σ’ ένορκόν γ’ άξιώ θέσθαι, τέκνον.
(‘Phil.: But, I implore you, do not leave me alone. Neopt.: Be assured, we will
stay. Phil.: Is it true that you will stay? Neopt.: Know it for certain. Phil.: I
really do not want to make you swear an oath, my son.’, S. P h . 809-11)

In view of the situation, Neoptolemus might, in Philoctetes’ opinion, very well expect
him to ask for an oath. By adding μήν ‘in truth’, ‘really’ Philoctetes anticipates this
expectation and, beforehand, he explicitly denies having this request for an oath. In
this way he makes it very difficult for Neoptolemus simply not to believe him.10
Often, as in (2), μήν anticipates a (possible) contrary or negative reaction on the part

7 Cf. Sicking (1986: 132-4); Sicking & Van Ophuijsen (1993); Wakkcr (1994: 307,343-64).
8 In the lyric parts of tragedy also Doric μάν is found, e.g. A. Ch. 963, P ers. 992; E. A le. 89.
9 For a comparable description see Sicking (1986: 132); Sicking & Van Ophuijsen (1993: 54-5).
10 Denniston (331) and Kamerbeek a.l. characterise this μήν as emphatic. Jcbb’s adversative para­
phrase T should prefer a promise on oath; however I do not like to ask for it’ seems far-fetched and
seems to be based on the assumption of an ellipse, which is entirely a d hoc.
214 GERRY WAKKER

of the addressee and explicitly contradicts it. This also explains why μήν is often
combined with ού, as in (2).11
Generally speaking, μήν is often used in clauses which express the contrary of
what (in the speaker’s opinion) the addressee supposes/wishes or might sup-
pose/wish (on the basis of what has been said before), cf. (3)

(3) Antigone: άλλ' όσης ό τόπος ή μάθω μολοΰσά ποι;

Oedipus: ναι, τέκνον, ε’ίπερ έστί γ ’ έξοικήσιμος.
Antigone: άλλ’ έστί μήν οίκητός'
(‘Ant.: But, in truth, shall Ig o and discover what place it is? Oed. Yes, my
child, at least if it is habitable. Ant.: But it really is inhabited’, S. OC 26-8)

With μήν Antigone explicitly marks the truth of her words, which is quite under­
standable since she expresses the contrary of what her father’s words imply. His
‘exclusively if ( - εΐπερ) it is inhabited/habitable’ implies that Oedipus is not sure
whether the place is inhabited.1112
By highlighting the truth of the proposition the speaker makes it virtually
impossible for the addressee not to believe him (unless, of course, the addressee
would explicitly say ‘Ido not believe you’, which, however, in normal conversation is
rather blunt and therefore rare). Given this semantic-pragmatic function it is not sur­
prising that μήν typically occurs in contexts where there is conversational interaction
between the interlocutors, i.e. in dialogic contexts, in which a speaker reacts on (the
possible implications of) the words of his addressee. Not surprisingly, within these
contexts μήν is typically found in declarative clauses, i.e. clauses that make an asser­
tion. The above examples, then, are illustrative of the typical usage of μήν.13

11 Cf. e.g. S. P h. 1273; E. A ndr. 256, Hec. 401, H e l 571, H era ct. 556.
12 Cf. e.g. alsoE. A le. 487 (if Diggle's textis correct, the speaker, Heracles, contradicts the possible
implication o f the chorus’s words, viz. that he should not try to perform his task); H e c . 401
(Odysseus contradicting Hecuba’s hope—as apparent from her words—that her child will not be
taken away from her); H era ct. 556 (Iolaus explicitly contradicting—the implication of — the girl’s
words, viz. that he might want to force her to die).
13 For μήν in other than purely dialogical contexts see section 3.1.5. In tragedy there are only a few
examples of μήν in what are, strictly speaking, non-declarative clauses (15 out of the 209 examples
studied), i.e. in interrogative (10 exx.) or directive (5 exx.) clauses. Moreover, o f these examples S.
E l. 973, for which see (24), is a rhetorical question which functions as a statement, i.e. as a declara­
tive clause. Also τ ί μήν; (S. A j. 668—τ ί μή; codd.— ; A. A . 672, E u . 203, S u p p l. 999) added to a
statement is a rhetorical means to underline the inevitability of the truth o f the previous statement:
‘what (else) then/in truth?’, ‘o f course’. It is also used to present a positive answer as inevitable (E.
R h . 706 τ ί μήν ού;) or, in a narrative, to underline the wholly natural character of the course of
action (E. R h . 955). In the three remaining questions (A. P ers. 233, E. A le. 518, E. R h . 175) μήν is,
in my opinion, added to the question to indicate that the speaker really wants the true answer, rather
than—as Dennislon (332) suggests— to add liveliness to the question, cf. A. P ers. 233: άλλά μήν
'ίμειρ' έμός παίς τήνδε θηρασαι πόλιν; ([Where is Athens? - far away -] ‘but - in all truth/I really
want to know the truth - did my son yearn to make this town his prey?’). In view o f the distance,
the queen finds it hard to believe that her son should be eager to capture Athens but she never­
theless wants to know the truth. One could say that with μήν the speaker marks his special interest
in the truth of the answer, even though this truth may be incredible or unexpected, or, in the case of
 EMPHASIS AND AFFIRMATION 215

3.1.1. Assenting κ α ν μην

In su c h d ia lo g ic c o n te x ts in trag ed y th e re a re m a n y e xam p les o f th e c o m b in atio n κ α ί

μ η ν , w h e re μ η ν m a y b e a r th e sa m e a ttitu d in a l v a lu e as in th e a b o v e ex a m p le s; b y
m e a n s o f κ α ί, w h ich , r o u g h ly sp eak in g , in d ic a te s th a t w h at fo llo w s sh o u ld b e i n c o r­
p o r a te d in to th e sa m e te x tu a l u n it as w h a t p re c e d e s,14 th e u tte ra n c e is clo sely lin k e d
to th e p re v io u s o n e , a lth o u g h i t is sp o k e n b y a d iffe re n t s p eak er. T h is is n o t s u rp ris­
in g sin c e th e κ α ί μ ή ν - c la u s e is a lw ay s a r e a c tio n to a p re v io u s q u e stio n , re q u e s t o r
sta te m e n t. In th a t c o n n e c tio n ( κ α ί) th e s p e a k e r p u ts fo rw a rd a p o in t w h ic h h e a b so ­
lu te ly w a n ts to stic k to, a p o in t o f w h ic h h e a s it w ere p erso n ally gu aran tees th e tru th
(μ η ν ). T h is p e rs o n a l g u a ra n te e m a y b e g iv e n fo r tw o (w id e ly d iv e rg e n t) re a so n s.
F irs t, th e s p e a k e r m a y e x p re s s his a sse n t a n d d o so in a r e a ssu rin g w ay , b e c a u se in
h is e y e s th e ad d re sse e p ro b a b ly w o u ld n o t e x p e c t this assent, cf. (4) a n d (5). A ltern a­
tiv ely , a s p eak er m a y w a n t to g uarantee th e tru th o f his statem en t b ecause h e explicitly
co n tra d ic ts th e a d d re sse e o r h is (su p p o sed ) ex p ectatio n s o r w ish e s, cf. (8) an d (9).

(4) Elcctra: ά λ λ ' ή ν έφ ης μ ο ι, τ ο ΰ τε θ ν η κ ό τ ο ς θ' ϋ π ε ρ

λ ε ξ α ιμ ’ α ν ό ρ θ ώ ς τ η ς κ α σ ιγ ν ή τ η ς θ' όμ ο ΰ .
C lyt.: κ α ί μ ην έ φ ίη μ ’·
(Έ 1 .: B u t i f y o u allo w m e, 1*11 sp e a k in a ju s t w ay o n b e h a lf o f the d e a d m a n
an d o f m y sis te r a lso . C lyt.: I n d e e d I a llo w y o u ’, S. E l. 554-6)

G iv e n the se rio u sly tro u b le d re la tio n sh ip b e tw e e n m o th e r a n d d a u g h te r it is n o t s u r­

p ris in g th a t C ly ta e m n e stra e m p h asises th e sin cerity/truth o f h e r p e rm issio n .15

(5 ) P en th e u s: ξ υ ν έ θ ε σ θ ε κ ο ιν ή τ ά δ ’, ϊ ν α β α κ χ ε ύ η τ ' ά ε ί.
D io n y su s: κ α ί μ η ν ξ υ ν ε θ έ μ η ν τ ο ΰ τ ό γ ’, 'ίσ θ ι, τω θ ε ω . 16

ού μήν (‘surely n o t...’), surprising or even unhoped for. For the latter, cf. E. Ale. 518, where Hera­
cles (after having heard that Admetus’ parents are still alive) rather incredulously asks ού μήν γυνή
γ' δλωλεν "Αλκηστις σέθεν; ('Surely your wife Alcestis has not gone?’), cf. also E. Rh. 175.
The directive öpa γ ε μήν (S. OC 587) does not function as an order that has to be performed
on its own, but serves rather to draw the attention to the next statement, which is in explicit con­
trast with the statement o f the previous speaker. Μήν has thus its normal attitudinal value. Given
the occurrence here o f γ ε μήν plus imperative it is probably best to read όνα γ ε μάυ δόμος (rather
than δναγε μάν) at A. Ch. 963. Here too there is a contrast, which is explained afterwards (‘arise,
you house; too long you have lain prostrate on the ground’). Note that, unlike S. OC 587, A. Ch.
963 is a real directive clause. In my opinion, μήν is added to the imperative to intensify the order, in
that it insists on its realisation: ‘make sure that you really arise’ (cf. also A. Supp. 1018). Moreover,
the orders concerned are either orders of which the speaker may doubt whether their addressee is able
to perform them (A. Ch. 963), or orders of a concessive nature that are not expected by the addressee
(A. A . 931; E. Cyc. 176); in these cases, μήν is added to underline that the speaker truly wants the
orders tobe executed, impossible or unexpected as this may seem.
14 Cf. Bakker (1993: 280, 284-91); Sicking & Van Ophuijsen (1993: 10-7); Sicking (1994: 22).
15 Cf. A. Supp. 468; S. Aj. 539,794, 990, Ant. 221, El. 1045, OT 345,749, 836,1004,1005; E. El.
648,661,670,1119.
15 For a discussion of the textual problems and the arguments in favour o f this text and interpreta­
tion I refer to Rijksbaron (1991: 100-1).
216 GERRY WAKKER

(‘Penth.: You have made this compact, so as to revel/celebrate the mysteries

of Bacchus for ever. Dion. I have, indeed, made this very compact—you can
be sure—with the god', E. B a . 807-8)

Pentheus does not expect Dionysus to agree with him. However, Dionysus does so,
at least partly: Ί have, indeed, made this compact—you can be sure. ’ With τφ θεφ,
however, he corrects Pentheus’ assumptions that he has made an arrangement with
the Theban women (cf. the apt remarks of Tyrrell a .l.) .

3.1.2. Excursus: κ α ί μην versus κα ί δή

It is exactly on this point, viz. that the addressee may not expect the assent, that these
examples of καί μην seem to differ from καί δή in, at first sight similar, assenting
contexts. I agree with Denniston’s remark, quoted in section 1, about the problems
involved in distinguishing μήν from δή. In fact, in many passages the difference is
not immediately obvious; yet both particles have, I submit, in their primary function
as attitudinal particle a semantic value of their own. A thorough comparison of the
two particles falls outside the scope of this paper, but, briefly stated, one may describe
the particles as follows. I have argued elsewhere (Wakker 1994: 351) that the pri­
mary meaning of δή is that of an attitudinal particle which demands the addressee’s
special attention for the (interesting and important) proposition presented by the
speaker. In its primary function δή may be paraphrased by means of such expres­
sions as French ‘voici’, Dutch ‘let wel’, ‘zie hier’, English ‘look, how interesting’.
When δή is used it is very difficult for the addressee not to pay attention to the
proposition, since he is specifically asked to do so. The meaning assigned to δή is, of
course, rather broad, and depending on the context δή may express various nuances
of this meaning (after all a speaker may have many reasons to ask for special atten­
tion).17 With μήν, as we have seen, a speaker confirms the truth of his proposition,
thus making it difficult for the addressee not to believe him. Consequently, in its pri­
mary function μήν prefers declaratives in dialogic contexts, while δή is found in all
types of utterance. Of course, there are also areas of overlap, e.g. in assenting con­
texts. While καί μήν is used to express assent whenever the addressee is likely not to
expect the assent, as we have seen above, καί δή merely expresses assent without any
idea of unexpectedness (‘and look’, ‘well’). In this combination κ α ί expresses,

17 To illustrate this with one example, from Classical Greek onwards δή is often found in conclu­
sions and other assertions which refer to observable, immediately understandable or generally known
facts. As elsewhere, in principle δή draws the addressee’s special attention to the (important and
interesting) conclusion presented, but since the addressee may as it were observe witli his own eyes
that the proposition is true, δή gels the nuance ‘obviously’, ‘as you see’, ‘clearly’ and the like, e.g.
in έστι δέ νέος· φοβούμεθα δή περ ί αύτώ, οΐον εΐκός π ε ρ ί νέω (‘he (= Cleinias) is young; and
so we are concerned about him, as is logical/to be expected about a young man’, PI. Eulhd. 275b2,
cf. S. A nt. 173, 939). I do not agree with Sicking (1986: 133-4, 136-7) and Sicking & Van
Ophuijsen (1993: 82-3, 85,141) that ‘obviously’ etc. is the general or basic meaning o f δή; in my
opinion, this semantic nuance is rather a later development of δή, which occurs only in specific
contexts.
 EMPHASIS AND AFFIRMATION 217

again, the close link with the previous utterance, whereas δή demands the addressee’s
special attention not so much for the assenting character of the statement as for the
fact that expressing the assent means at the same time performing the state of affairs
assented to, cf.

(6) Heracles: τόλμα προτείναι χέίρα και θιγεΐν ξένης.

Admetus: καί δή προτείνω,...
(‘Her.: Have the courage to stretch forth your hand and to touch your guest.
Adm.: See, I am stretching forth (my hand)’, E. A le . 1117-8)

(7) Electra: άλλ' ήν έφής μοι, του τεθνηκότος θ’ υπέρ

λέξαιμ’ αν όρθώς της κασνγνήτης θ’ όμοΰ.
Clyt.: καί μην έφίημ’1'ε ί δέ μ’ ώδ' άεί λόγους
έξήρχες, οΰκ αν ήσθα λυπηρά κλΰειν.
Electra: καί δή λέγω σοι,
(Έ1.: But if you allow me, I ’ll speak in a just way on behalf of the dead man
and of my sister also. Clyt.: Indeed I allow you. If you had always begun
your speeches in such a manner, you would not have been painful to listen to.
El.: And see I am speaking’, S. E l. 554-8)

In (6) Admetus’ reaction to Heracles’ encouraging words is not really unexpected.

With δή Admetus asks attention for the fact that he is already performing the state of
affairs concerned.18 Similarly, in (7) in using κ αί δή (‘and see’) Electra not only
expresses the acceptance of the permission, but at the same time she emphasises that
she is about to perform the state of affairs in question.19 As Denniston (251 iii)
observes, this καί δή is usually combined with a present or perfect tense, which is
fully compatible with the observation that δή demands the addressee’s special atten­
tion for the fact that the state of affairs in question is already (being) performed.

3.1.3. ‘Contradicting’ κ α ί μήν

To return to και μήν, the idea of stating something unexpected is all the stronger in

18 There is a varia lectio κ α ί μήν, which would imply that Admetus marks the iruth/sinccrity of his
assent, as i f Heracles would not perhaps expect him to assent. Given the fact, however, that in this
άγων Admetus every time obeys Heracles (cf. 1108,1110, 1112,1120) and that, therefore, it will
not really amaze Heracles that Admetus is performing this particular order, it seems more likely that
Admetus with δή underlines that he is already performing the state o f affairs in question. Some
other examples (where there is no v./.) of an order followed by a καί δή-clause marking the accep­
tance and performance of the state of affairs in questionare: A. Eu. 894, Supp. 438,507; S.Ani. 245,
EL 317, 892,1436,1464, OC 173, Ph. 818; E. Ph. 387.
19 Note in (7) the occurrence o f both καί μήν and κ α ί δή in the same passage. In my opinion they
cannot be exchanged here, κα ί μήν emphasizing that the assent may be unexpected, but is truly sin­
cere, κ α ί δή simply expressing the acceptance and the performance of the permission given in 556.
218 GERRY WAKKER

those examples where the speaker is actually contradicting the addressee or his sup­
posed expectations or wishes, cf. (8) and (9)20

(8) Antigone: καλώς συ μέν τοΐς, τοίς 5’ έγώ ’δόκουν φρονεΐν.

Ismene: καί μην 'ίση νφν έστιν ή 'ξαμαρτία.
(‘Ant.: Some thought you were right, and some thought I was. Ism.: And for
sure our offence is equal’, S. Ant. 557-8)

With her remark Ismene contradicts the implication of Antigone’s argumentation

(‘you are not guilty, you can live on’). Since she knows that Antigone does not agree
with her view, she underscores with μήν the truth of her words, thus making it all the
more difficult for Antigone not to accept this view.

(9) Prom.: άλλ' έκδίδάσκει πάνθ' ό γηράσκων χρόνος.

Hermes: καί μήν συ γ' οΰπω σωφρονεϊν έπίστασαι.
(‘Prom.: But ever-ageing time teaches all things. Herrn.: And truly, you at
least have not yet learned to keep a sober mind’, A. Pr. 981-2)

With μήν, Hermes underlines the truth of his statement, and with good reason, since
his statement is in clear contrast with the implications of the previous statement.21

3.1.4. ή μήν and the differences between ή and μήν

Another frequent particle combination in contexts in which the speaker believes his
statement to be in contrast with what the addressee, on the basis of what has previ­
ously been said, might suppose or wish, is ή μήν.
As noted in section 1, just like μήν, ή is generally described as an affirmative
particle. Two questions must now be asked: first, what is the difference between ή
and μήν and, second, what is the value of the combination ή μήν.
As to the value of ή, just like Sicking (sec section 1), I accept Ruijgh’s hypothe­
sis (1971: 191) that originally ή and ού were counterparts, the one underlining the
affirmative, th e o t h e r e x p r e s s i n g th e n e g a t i v e c h a r a c t e r o f th e utterance.22Accord­
ingly, the value of ή in declarative clauses may be paraphrased as ‘there is no deny­
ing’, ‘it is true that’. In other words, with ή a speaker expresses beforehand (ή
always occurs in front position) that th e r e is no denying that his proposition is really
true. If this hypothesis is correct, it implies that in declarative clauses, ή, just like où,

20 Not surprisingly, in such contexts καί δή is not found.

21 Cf. A . A . 1254, S .A j. 531, Anf. 1054, £.7. 321, 1188, O C 396, O T 987,1066;E.A/c. 105,1099,
A ndr. 81, H e l 554, IA 20, Io n 985, Or. 109.
22 Although ή and ού are, I submit, counterparts, their functions are not symmetric. While ή rein­
forces the truth value (‘it is true’) o f the utterance, a truth value the utterance would have had any­
w a y, also without ή, the presence of ού is essential to express the truth value ‘it is not true that’;
without ού a different world would have been represented.
 EMPHASIS AND AFFIRMATION 219

functions at the representational level of discourse.23 Μήν, however, as I have argued

above, functions at the interactional level of discourse. This difference in discourse
level may explain the following distributional differences.

1. As one would expect on the basis of the hypothesis that η and ού originally were
counterparts, ή is never combined with ού, unlike μήν, which is often thus combined,
cf. (2) and n. 11.

2. Whereas μήν is hardly ever found in real questions (cf. n. 13), ή is often used in
this way. With an ή-question the speaker asks for an explanation of something just
said or implied, or for confirmation of something that has been said, implied, or sug­
gested by the previous speaker. The tone is often urgent, the speaker expressing his
surprise, incredulity, anxiety, eagerness, indignation etc. about the information just
heard and verifying whether the inference he is drawing is justified or not: ‘is it true
th at...?’ > ‘does this mean t h a t ‘What, do you mean to say that ,..?’24*

(10) Electra: ποίω; τί φωνεΐς; Orestes: οΰ τό Φωκέων πέδον

ύπεξεπέμφ&ην σή προμηθίςί χερόίν.
Electra: ή κείνος ούτος ον ποτ' έκ πολλών εγώ
μόνον προσηΰρον πιστόν ...;
Orestes: οδ' έστί'
(‘EL: W hatman? What are you saying? Or.: The man by whose guidance I
was conveyed through your foresight to the Phocian plain. El.: Is this really
the one out of many whom I found loyal once? Or.: This is he’, S. E l. 1349-
53)25

Electra, who has failed to recognise the man concerned, is highly surprised about the
inference she has to draw from Orestes’ words.
Sometimes an ή-qucstion is preceded by πώς ειπας; etc., which asks for prelimi­
nary enlightenment, cf.
(11) Ιο: τίς ούν ό λύσων έστίν ακοντος Διός;
Prom.: των σών τιν' αύτόν έκγόνων είναι χρεών.
23 I would argue that this holds at least for ή and ού in declarative clauses. As yet I am not sure
whether·?! and ού introducing questions function at the representational level, too, or whether they
function rather at the interactional level. If the latter is the case, the speaker expresses his attitude
towards the truth of the state o f affairs concerned: by the use of ή he indicates that he wants some
piece o f information confirmed as undeniably true (cf. [10]-[11]); by the use of ού the speaker indi­
cates that he himself believes the proposition to be true, and expects the addressee to have the same
attitude towards this proposition, e.g. ούχ ούτως έλεγες; (‘this is what you said, isn't it?’, PI. R.
324b6). However, this subject falls outside the scope o f my paper.
24 In principle neutral questions are concerned, to which, depending on the context, the answer may
be ‘yes’ or ‘no’, cf. for ‘yes’ e.g. (10), (11), E. ion 990, IT 741, Supp. 935, for ‘no’ E. Hipp. 1448,
Tr. 250. It would be interesting (but this falls outside the scope of my paper) to study the dif­
ference between questions introduced by fl, ή καί, ή γάρ and those introduced by Spa. For ή γάρ-
uestions see Van Erp Taalman Kip (this volume).
Ì 5 Cf. e.g. A. Th. 183; S. OC 2 6 ,3 0 ,Ph. 810; E. J T 1176.
220 GERRY WAKKER

Ιο: πώς εΐπας; η 'μός παίς σ’ απαλλάξει κακών;

(‘Ιο: Who then is to loose you against the will of Zeus? Prom.: It is to be one
of your own lineage. Io: How do you say? Do you mean that a child of mine
will release you from your misery?1, A. P r . 771-3)26

By using ή Io underlines that she can hardly believe and understand what
Prometheus is saying and that, for that reason, she is checking whether she has drawn
the right conclusion.

3. Whereas with μην, as has been argued above, the speaker as it were p e r s o n a l l y
guarantees the truth of the proposition concerned and generally reacts to, and
expresses the contrary of, what the addressee might either suppose or wish, ή (as a
representational particle) is as it were more objective ‘there is no denying’, i.e. neither
for the speaker nor for the addressee: ‘there is no denying—whether or not you and I
think it unbelievable or undesirable.’27 By ή, the speaker presents the proposition as
u n d e n i a b l y true. Often, of course, a speaker will underscore the truth of his proposi­
tion in this way when he supposes the addressee to be inclined to disbelieve him (in
this respect η resembles μην). However, an ή-clausc does not necessarily react to
information given previously by the addressee, as may be inferred from examples
where η is used by a character entering upon the stage without having heard the pre­
vious words, cf.

(12) Creon: ή πόλλ' έπηλθον είσιδείν χρήζων σ’, αναξ

Έ τεόκλεες... I ...
Eteocles: καί μην έγώ σ' έχρηζον είσιδέΐν, Κρέον
(‘Cr.: Truly, I have gone to many places, hoping to see you, king Eteocles.
Et.: And I really wanted to see you’, E. P h . 697-700)

In 690 Eteocles wants to send forth someone to fetch Creon, unnecessarily, since at
that very moment Creon appears, opening his words with ή. Note the opposition with
καί μην used by Eteocles in his reaction. With κ α ί μήν Eteocles underscores the
truth/sincerity of his assenting words, from which we may infer that he thinks it
unlikely that Creon expects that he (E.) wants to see him (= Creon).
Generally speaking, η is used in all kinds of opening words which do not react to
the words of a previous speaker, e.g. in the closing lines of the tragedy spoken by the
chorus (S. A j . 1418), in the opening line of a long speech or prophecy which has
been announced in advance, or in the opening line of a choral song (A. E u . 34, P e r s .
852, P r . 887; E. A n d r . 274, M e d . 579). Μήν is never used in this way.
4. The above description, notably the idea that ή functions at the representational level
and forces the truth of something both upon the speaker and upon the addressee

26 Cf. e.g. S. EL 407, OT 622,943, Tr. 977.

27 The observations in Sicking (1986: 133, 137) point in the same direction. Unfortunately, he
does not further investigate this point in Sicking & van Ophuijsen (1993: 54-7).
 EMPHASIS AND AFFIRMATION 221

(‘there is no denying, whether or not you and I think it believable’) also explains its
frequent collocation with a number of attitudinal particles which are not found in
combination with μην. The strong personal conviction conveyed by μην apparently
clashed with the values of these altitudinal particles. We find the following combina­
tions with ή (and not with μην).

- ή dpa or η τάρα (= τοι dpa). As stated in section 2, the attitudinal particle dpa
expresses the idea that the speaker cannot but accept the truth of the proposition
involved, although he is surprised at it. He thus anticipates or wants to elicit a possi­
ble feeling of surprise on the part of the addressee. Ή (τ)άρα is often found in pas­
sages in which a speaker draws a conclusion, a conclusion that is inevitable in view of
the previous words or of the situation and that at the same time is surprising—often
disappointing—for the speaker (and possibly also for the addressee). An example is
(13), where in a long rhesis Admetus is reproaching his father that he has not shown
the courage to die in his place. In his own words, ‘put to the test, you showed who
you are and I do not reckon myself your true-born son.’ Admetus then continues
with

(13) η τάρα πάντων διαπρέπεις άψυχίςι,

δς...
(‘In all truth [η], I tell you [τοι], you are apparently [dpa] without peer
among men in cowardice, you, who ...’, E. A le . 642)28

- ή δή, too, may be used in conclusions. TH indicates that the conclusion cannot be
denied, whereas with δή the speaker asks special attention for it, mostly because he
feels indignant about it. This combination occurs both in declaratives (A. C h . 742; E.
S u p p . 423) and in concluding, checking questions, cf. (14), where Chrysothemis has
just informed Electra that if she will not stop lamenting she will be put away into a
dark cave. Surprised and indignant, Electra reacts as follows:

(14) ή ταΰτα δή με κ α ι βεβούλευνται ποιείν;

(Tn all truth [ή], is that really [δή] what they have decided to do to me?’, S.
385)29
E l.

- With ή που the speaker expresses the idea that there is no denying that the proposi­
tion X is true (ή), but at the same time που (‘perhaps’) mitigates the affirmative tone,
giving the declarative the character of a supposition, of a guess that one believes to be
true ‘surely, I suppose’, cf. (15).

(15) Aias (in a rhesis): ή που τάλαινα, τήνδ’ όταν κλυη φάτιν,
ήσει μέγαν κωκυτόν έν πάση πόλει.

28 C f.e.g.E .A te. 732,Andr. 274, Hec. 1118,Heracl. 651, Hipp. 480, M 1330.
29 Cf. e.g. E. Rh. 686.
 GERRY WAKKER

(‘Poor woman, when she hears this news she will surely, I guess, utter a loud
wailing in all the city’, S. A j. 850)

The above four observations all point to a description along the following lines: η is a
representational particle; by using ή a speaker expresses beforehand (independently
of previous information) that t h e r e is n o d e n y i n g that his statement is r e a l l y true
(whether or not you, addressee, and I, speaker, will think this undesirable or unbeliev­
able). With the attitudinal particle μήν, on the other hand, the speaker as it were per­
sonally guarantees the truth of the proposition, often because he supposes his
proposition to be in contrast with what the addressee supposes or wishes.

Now to return to ή μήν, this combination yields the following value: ‘there is no
denying, and I am myself very much convinced—in contrast to what you may sup­
pose or wish—t h a t i t thus makes the declarative have the value of a s t r o n g a s s u r ­
a n c e , the truth of the declarative being affirmed both ‘objectively’, at the representa­
tional level, and ‘subjectively’, at the interactional level, cf.

(16) Creon: ... ήν σ’ ελω ποτέ -

Oedipus: τις δ' άν με τώνδε συμμάχων ελοι βίςι;
Creon: ή μήν συ κάνευ τοΰδε λυπηθείς έση.
(‘Cr.: If I ever catch you. Oed.: And who could catch me against the will of
these allies? Cr.: I swear that in truth even without that happening you shall
suffer pain’, S. O C 814-6)30

With ή μήν Creon, beforehand, emphatically contradicts Oedipus’ possible expecta­

tions, viz. that nothing will happen to him against the will of his friends. Note that
Creon’s words have a threatening tone; this threatening nuance, however, does not
belong to the semantics of μήν, but arises out of the context.
The supposed negative reaction may also be less explicitly present in the text,
notably in the well known use of ή μήν in oaths, ή μήν ‘you, addressee, perhaps you
will not believe me, but, there will be no denying, I assure you...’, cf.

(17) ομνυσι δ’ ... I ... I ή μήν λαπάξειν άστυ Καδμείων βίςι I Διάς'
(‘He swears an oath that verily he will lay waste the city of the Cadmcans in
spite of Zeus’, A. T h . 529-32)31

The combination of a subjective (‘I assure you’) and an objective (‘there is no deny­
ing’) guarantee that the proposition in the (near) future will be true is, of course, very

30 Cf. c.g. A. A. 1098, Pr. 73,167,907; E. Ale. 64.

31 Cf. e.g. S. Tr. 1186,1187; and in indirect discourse S. Tr. 256, Ph. 593; E .JA 475. Quite often
these oaths are presented in indirect speech dependent on ‘I swear’ (performative verb) or ‘he
swears’, cf. (17); cf. Dennislon (351).
 EMPHASIS AND AFFIRMATION· 223

appropriate in pledging oaths. The speaker does not want the addressee to have any
doubt aboutit.

3.1.5. Altitudinal μην in non-dialogic contexts

All examples up to now occur in dialogic contexts, where an addressee is present who
really participates in the conversation, i.e. speaker and addressee alternatively take
turns in the conversation. Μήν, which by its very meaning prefers contexts in which
an addressee is actively present, also occurs, however, in monologic contexts, i.e. in
narrative, descriptive and argumentative contexts. In tragedy, these are contexts where
one of the characters functions as it were as the reporter/narrator who phrases the
text, and where the (internal) addressee is not actively involved in the communica­
tion.32 In such narratives or descriptions, with μήν the character-narrator anticipates
(and contradicts) the possible conclusions that his addressee(s) may draw from the
facts presented earlier. In many of these examples we find the combination ού μήν,
cf. (18), but there are also examples not fitting this pattern, cf. (19).

(18) où μήν ΐσην γ' ετεισεν, άλλα ...

(‘He really did not pay an even penally’, S. O T 810)33

Having heard the story of how Oedipus was hit on his head the addressees, Iocaste
and the chorus, might perhaps suppose that Oedipus was no longer able to act or that
he would merely do the same as his adversary. The où μήν-clause explicitly contra­
dicts this expectation.34

(19) ε ΐε ν τίν' αρχήν πρώτα σ' έξείπω κακών,

ποίας τελευτάς; τίνα μέσον τάξω λόγον;
και μήν δι' ορθρων γ' οϋποτ’ έξελίμπανον
θρυλοΰσ' ά γ’ είπεΐν ήθελον κατ' ομμα σόν,
ε ί δή γενοίμην δειμάτων έλευθέρα
των πρόσθε.
( ‘So be it. Where s h a l l I begin th e h a r s h w o r d s I have to say? Where end?
Which words shall I put in the midst? And truly dawn after dawn I never
failed to repeat over and over what I wanted to tell you to your face if ever I
was free from past terrors’, E. E l. 907-12) ·

Here the speaker contradicts the possible implications of her own questions (907-8),
viz. that she has never thought about the things she would say.

32 See De Jong (1991) for a discussion of one type of narrative in drama: Euripidcan messenger-
speeches.
33 In many examples, as in (18), γε is added in the μήν-clause to explicitly make this (usually
unexpected and therefore most important) constituent the scope of the utterance, cf. also (2), (9),
(19), (20), (25).
34 Cf. e.g. A. A . 1279, P r. 268; E. R h. 778,958.
 GERRYWAKKER

Interestingly, in monologic argumentative contexts dialogic phenomena may occur

(such as questions or directives, presupposing the involvement of an addressee; inter­
jections; the presence of first and second person pronouns etc.)·35 In such contexts
μήν is also found, cf.

(20) ού μήν έρεΐς γέ μ’ ώς άτιμάζοντα σόν

γήρας θανεΐν προύδωκας,...
(‘Surely you will not say that you have given me up to die because I failed in
respect for your age’, E. A l e . 658-9)

Admetus is summing up the arguments as to why his father should have chosen to
die in his place: he has got everything during his life (long life, kingship, son). With
the ού μήν-clause Admetus explicitly contradicts the possible counter-argument
Pheres might think of: ‘all right, during my life I have had all, but you did not honour
me.’36
As the above examples show, μήν in non-dialogic contexts has the same function
as in dialogic contexts, the only difference being that by definition the
speaker/narrator cannot react to (the implications of) his addressee’s previous words.
He rather contradicts the possible conclusions his addressee may draw from the facts
that the speaker himself has presented just before.

3.2. T h e so-called ad v e rsativ e value of μήν

In many of the above examples, where μήν corrects or eliminates the (implications of
the) previous proposition(s), an ‘adversative’ nuance is present as well, cf. exx. (2),
(8), (9), (18), (19), (20). It seems unwarranted, however, to attribute in these cases an
a d v e r s a t i v e c o n n e c tin g f u n c t i o n to μήν.37 The adversativity results from the fact that
contrasting assertions are made and does not as such belong to the meaning of μήν.
A clear case in point is

(21) Creon: ού βούλομαι τον μάντιν άντειπειν κακώς.

Teiresias: και μήν λέγεις, ψευδή με θεσπίζειν λέγων.
(‘Cr.: I do not wish to reply rudely to the prophet. Teir.: And truly you do
speak rudely, saying that my prophecies are false’, S. A n t . 1053-4)

Μήν, as elsewhere, means ‘and truly (you do that)’; by the use of μήν Teiresias
insists on the truth of what he says, clearly because he is contradicting Creon. The
adversative relationship is not explicitly indicated as such, however, and μήν has its
normal function as an attitudinal particle at the interactional level, not as an adversative

35 Cf. Kroon (1995: 111-5).

36 Dcnnislon (338) attributes a progressive function to this μήν ‘nor again’, unnecessarily, in my
opinion. Comparable examples are A . A . 1068; S. El. 817; E. Ale. 653,1018; Med. 1032.
37 In (3), this is impossible anyhow, given the presence of αλλά, but also in other cases the adver­
sativity is not part of the semantics of μήν.
 EMPHASIS AND AFFIRMATION 225

connector at the representational level. Otherwise stated, μήν does not itself express
the adversative relationship, but by its very meaning it is very much compatible with
such a context.
For Homer, as I stated in section 1, it has been argued that μήν never fully has
the status of an adversative connector. Everywhere its atdtudinal value applies. It must
now be asked whether the same holds for tragedy in those cases where a nuance of
adversativity is present, at least in the context, and where commentators and/or Den-
niston attribute an adversative value to μήν.38 For most combinations (άλλα μήν, καί
μήν, κα ί μήν ... γε, άλλ’ ού μήν ... γε) the same description might be given as for
(21) : the adversativity results from the fact that contrasting assertions are made and
does not belong to the meaning of μήν.
Somewhat problematic is a group of eight examples of γ ε μήν in Aeschylus and
Euripides, where usually the combination γε μήν as a whole is given an adversative
value. What all examples have in common is the fact that they occur in a monologic
context (at any rate they do not react directly to the words of the addressee that pre­
cede) and that there is a strong verbal contrast, cf.

(22) ούκ οΐδ’ ακριβώς; είκάσαι γε μήν πάρα.

( Ί don’t know certainly; it is, to be sure, possible to divine’, E. R h . 284, cf.
A. A. 1378; E. E l. 754, H ip p . 1340, O r . 1083)

(23) αϋτη κατ’ Ά ργος βασιλικόν τέξει γένος,

μακροΰ λόγου δει ταΰτ’ έπεξελθέίν τορώς.
σπόρος γε μήν έκ τησδε φύσεται θρασύς
τόξοισι κλεινός, ός πόνων έκ τώνδ' έμέ
λύσει.
(‘she it is that shall give birth in Argos to a royal line—to set this forth at
large needs a long discourse—at any rate, her seed, to be sure, shall be bom a
man of daring, renowned with the bow, who shall deliver me from these toils’,
A. P r . 869-73; cf. E u . 51, T h . 1062)

Dcnniston (348) regards the combination γ ε μήν as one particle with adversative
connecting force. It is doubtful, however, whether this is necessary in all cases pre­
sented by Dcnniston. In many of his examples, just as in (22) above, the constituent
with γε is in explicit contrast with a constituent in the preceding context. In fact, γε is
often used to focus the attention on a constituent that is in (implicit or explicit) con­
trast with some other constituent (cf. Denniston 115-9; Wakker 1994: 308-10). In
cases like (23) there is some contrast as well: (23) occurs in a long speech by
Prometheus containing a narrative about lo’s wandering as well as a prediction about
the future. Since this prediction would need a long story to be told in all details, only

38 Such as, for instance, in A. A. 1068,1254,1279, Pr. 268,982,985; S. Aj. 531, Am. 558, El. 321,
817, 1188, OC 396, OT 810, 987,1066, Ph. 1273; E. Ale. 105,1018,1099, Mel. 308, Heracl. 556,
885, Ion 985.
226 GERRY WAKKER

the essence is presented. There is thus a contrast between μακροΰ λόγου and the
summary presented in the γε μήν-clause, cf. Griffith’s paraphrase a d lo c . ‘but (this)
at any rate (I will tell you)’. If we assume that in all eight γε μήν-cases γ ε focuses
the attention on the contrast, this leaves us with the question as to which value is
expressed by μην. Since contrasting states of affairs are presented a nuance of adver-
sativity is inherently present; μήν could be said to mark this relation. But it certainly
also fulfils its primary function of an attitudinal particle: the addressee is perhaps not
likely to accept the truth of the statement concerned. Μήν functions, then, primarily as
an interactional particle, though it functions perhaps also as a connector, marking the
relation of adversativity at the representational level of discourse. Note, however, that
we are dealing here with a group of only eight examples, a small minority of all cases.
Be that as it may, in all supposedly adversative examples μήν has (also) its primary
attitudinal value. There are no examples where μήν must be taken as just an adversa­
tive connector.

3.3. The progressive value of μήν

Just as μήν ‘truly’, ‘really’ is highly appropriate in adversative contexts, it is also

very much at home in enumerations, to mark an item of which the speaker may expect
that it will elicit the addressee’s disbelief or surprise. Anticipating a reaction of dis­
belief he marks the truth of what he is presenting with μήν. We thus often find μήν in
a climax, cf.

(24) πρώτον μεν ευσέβειαν ... I ... ο'ίση... I

έπειτα S '... έλευθέρα I καλή ...καί γόμων επαξίων 1
τεύξη- 1λόγων γε μήν εύκλειαν ούχ όρςίς όσην I
σαυτη τε κάμοί προσβάλεις...
(‘first you will win praise of piety; next you will be called free and you will
obtain a worthy marriage. Then, don’t you see what fair fame you will win for
yourself and for me?’, S. E l. 968-74)

Electra is sketching the positive consequences that Chrysothemis will experience if

she follows her advice. The mere number of the lines devoted to each item (2, 3 and
12, respectively) indicates that the third item is characterised as a climax. Just as in
(22) above, γε may be used to focus the attention on a contrast. Electra’s λόγων
contrasts with the έργα (freedom and marriage) implied by the preceding context.
Μήν is added because in the circumstances Chrysothemis might perhaps not think of
this point: ‘Really, how much εύκλεια you will have.’ While the attitudinal value
might, then, be present, μήν is used in a totally different type of discourse
(argumentative monologue). Thereby it acquires a progressive nuance, since it intro-
 EMPHASIS AND AFFIRMATION 227

duces, at the presentational level of discourse, a new point, that the addressee proba­
bly did not expect39 Compare also, in a dialogic context

(25) και μήν τά γ' αλλα κωφά και παλαί’ έπη

([The chorus advise Oedipus to consult Teiresias; Oedipus replies that he has
already sent for Teiresiasl Ch. ‘All the rest, to be sure, is vague and ancient
rumour’, S.OT290)40

Having made their point about consulting Teiresias concerning the murder, the
chorus put forward a new point in connection with the murder, namely the various
stories that are being told. Beforehand, they indicate that these stories are definitely
vague and old. The chorus add μην to prevent Oedipus from estimating the impor­
tance of the stories in a different way (which he does indeed, cf. 291). Here, too, μήν
seems to have mainly an attitudinal value. The effect of progression is certainly pre­
sent, but arises out of the context.41
There are also examples in narrative contexts, e.g. within a messenger’s speech

(26) The Greeks have encouraged themselves with the following words: ‘Free
your native land, your children, wives, gods, tombs of your ancestors. Now
you fight for your all’
και μήν παρ' ήμών Περσίδος γλωσσης ρόθος
ύπηντίαζε, κοΰκέτ' ήν μέλλειν ακμή.
(‘And really, from our side a clamour of Persian cries answered, and the time
brooked no delay’, A. P e r s . 406-7)

This μήν is commonly taken as progressive, introducing a new point (Denniston,

Broadhead a .l., Sidgwick a .I.) . It is, however, more than that. After the description of
the way the Greeks prepared themselves, the adressees might perhaps infer that the
Persians did not react anymore. With και μήν this possible inference is explicitly
contradicted. Once again, then, μήν has primarily an attitudinal value: the progressive
nuance is due to the context.42

A transitional stage between interactional (i.e. attitudinal) and presentational (i.e. pro­
gressive) μήν is probably to be seen in the use of καί μήν (which is characteristic of
drama) to mark the entrance of a new character on the stage (27) or as a marker of the
sudden realisation of some fact (28):

39 Cf. with γ ε μήν E. Rh. 196; with και μήν ... (γε) A. A. 1188, Ch. 205, Th. 668 (ουδέ ... μήν);
E. Andr. 672, Hec. 317,824.
40 Cf. E. AL·. 516 (the new item πατήρ γ ε μήν is in contrast with παΐδες; at the same time the
proposition contradicts the possible idea of the addressee that the death of his father is a very
serious matter), E. Cyc. 141, Hel. 1071,1079; A. Ch. 174.
41 Denniston (354) and Dawe a.l. claim that μήν here expresses ‘agreement’ or ‘assent’: 'aye, truly’,
‘well certainly’. This seems rather improbable.
42 Some other examples: A. Pr. 459, Th. 538.
228 GERRY WAKKER

(27a) καί μην πρό πυλών ήδ' Ισμήνη (S. A n t. 526)

(27b) δδε μην Αίμων (S. A n t. 626; the only example of μήν alone)
(27c) και μήν όρώ... Εύρυδίκην (S. A n t. 1180)
(27d) καί μήν οδ' άναξ αυτός έφήκει (sc. Creon; S. A n t . 1257)

(28) Admetus: ώ δαΐμον, οϊας συζύγου μ' άποστερείς.

Alcestis: και μήν σκοτεινόν ομμα μου βαρύνεται.
Admetus: άπωλόμην αρ’, ε ϊ με δή λείψεις ...
(‘Adm.: Oh my destiny, from what wife are you bereaving me’ Al.: Ah! my
eyes grow heavy and dark. Adm.: I am lost, then, if you will really leave me’,
E. A l e . 384-6143

These two groups of examples have in common that they do not express a direct
reaction to what precedes. However, they do draw attention to an event that will not
have been expected at that very moment by the addressee. In this sense, then, a new
point is involved. When a new character is introduced in a και μήν-clause, this char­
acter is always a person whose entrance has not been announced in the preceding
lines. Kai μήν thus expresses, at the interactional level, ‘really there we have’ (the
speaker expresses his surprise that al that very moment... is coming and wants his
addressee(s) to share that feeling). On the other hand, at the presentational level, μήν
highlights a—surprising—new turn in the course of the events/discussion. Concern­
ing (27b), Denniston (356 (6)) supposes that μήν has a breaking-off function, but
this term is more suited for cases like (29), where the combination άλλα ... γάρ has
an explicit breaking-off function:

(29) άλλ' οδε γάρ δή βασιλεύς χώρας, I Κρέων

(‘but enough about this for [άλλα ... γάρ] here comes - please note [δή] - the
king of the land, Creon’, S. A n t. 155-6)

the same applies to cases like (28). The state of affairs described in
M u ta tis m u ta n d is
the καί μήν-clause is new and unexpected and constitutes at the same time a turn in
the course of the events.
I should add that καί δή, too, is used when a new character is entering on the
stage. Just as in (4)-(5) as opposed to (6)-(7), the difference between καί μήν and
καί δή seems to be that the former is connected with unexpected events, while the
latter is connected with events that are expected, cf.

(30) Medea: π άλαι... I καραδοκώ τάκείθεν οι προβήσεται.

καί δή δέδορκα τόνδε των Ίάσονος
στείχοντ' οπαδών

43 Cf. E. A ndr. 820, Β α. 918,957, Ιοη 201, I T 1050.

 EMPHASIS AND AFFIRMATION 229

(‘Long am I expecting what from there shall befall. And look, I see a man
from Iason’s servants coming here’, E. M e d . 1116-9)44

As in (6)-(7), with κ α ί δή the speaker asks attention for her (interesting) statement;
and, in fact, precisely at that moment the event which is being talked about takes
place.

4. Concluding remarks

In summary, μήν in tragedy may be described as follows. In most cases μην func­
tions at the interactional level o f discourse, as an attitudinal particle; by the use of μήν
the speaker insists on the truth o f his proposition, in this way anticipating a negative
reaction o f the addressee, who on the basis of previous information or of the situation
in general might not expect this proposition to be true. Not surprisingly, most cases
of μήν are found in declarative utterances in dialogic contexts.
By its strongly affirmative (and corrective) value μήν is especially at home in
adversative contexts: it corrects or eliminates the previous statement o r its implica­
tions. However, in tragedy, μήν never develops into a purely connective adversative
particle at the representational level. In all cases it functions at the interactional level.
On the other hand, the strongly affirmative value o f μήν makes it a suitable
means to highlight a surprising item, or a climax in enumerations. On the basis of
this use we may explain its progressive function at the presentational level, where it
serves to introduce a new step in the narrative or argumentation or to mark a sudden
turn in the course of the events. In tragedy this use of μήν is found when μήν marks
the entrance o f a new character upon the stage. The attitudinal value does not seem
wholly lost, however, as is apparent from a comparison with κ α ι δή in p a r a lle l con­
texts.
By the above description o f μήν I hope to have shown the advantage o f using
functional criteria such as the level of discourse, type o f context, type o f utterance,
relation speaker - addressee. These criteria made it possible to provide a description
of the uses o f μήν in tragedy which is more consistent than the one presented by
Denniston. M oreover these criteria made it possible to distinguish μήν from other
particles characterised by Denniston as ‘em phatic’ and/or ‘affirm ative’. Γ ε is a
scope particle with limitative value which functions at the representational level. It is
thus rather different from μήν both as to value and as to the discourse level with
which it is connected. TH primarily functions at the representational level, too; in
using ή the speaker insists on the truth o f a proposition in a more or less objective
way (‘there is no denying’). In other words, in using ή a speaker forces the truth of
his utterance upon the addressee; he presents this truth is inevitable, whether or not
speaker and addressee think it unbelievable o r undesirable. In using the attitudinal
and interactional μήν, on the other hand, a speaker as it were personally guarantees

44 Cf. e.g. S. Aj. 544: E. Cyc. 488, Supp. 1114.

 GERRY WAKKER

the truth of his statement and anticipates a possible negative reaction of the addressee.
Δή and μήν, finally, are both interactive particles, but the value of δή is widely differ­
ent from that of μήν: δή asks attention for the (interesting) proposition presented.
Since the reasons for doing so may be manifold, δή figures in many types of
contexts, whereas the use of μήν is much more restricted. On the basis of these
(rather) global semantic values it was possible to explain a number of distributional
features of the three particles μήν, ή and δή in tragedy (e.g. the fact that ή άρα does
and *μήν σρα does not occur), as well as some differences in nuance when they
occur in similar contexts, cf. και μήν and και δή in assenting contexts and their use
to mark the entrance of a new character upon the stage.45

References

Abraham (ed.), W.
1986 T i j d s c h r i f t v o o r T e k s t- e n T a a l w e t e n s c h a p ( i ' l l ) 6 no.2 (special issue on
particles)
Bakker, E.J.
1986 ‘οσπερ en ε'ίπερ: een aspect van Attische conversatie’. L a m p a s 19, 142-58
1988 L i n g u i s t i c s a n d F o r m u l a s i n H o m e r . S c a l a r i t y a n d t h e D e s c r i p t i o n o f th e
P a r tic le per. Amsterdam/Philadelphia: Benjamins
1993 ‘Boundaries, Topic and the Structure of Discourse: An Investigation of the
Ancient Greek Particle d e ’. S tu d ie s in L a n g u a g e 17, 275-311
Denniston, J.D.
19542 T h e G r e e k P a r tic le s . Oxford
Foolen, A.
1991 ‘Polyfunctionality and the semantics of adversative conjunctions’. M u l t i l i n ­
g u a 10-1/2, 79-92
1993 D e b e te k e n is v a n p a r tik e ls : E e n d o k u m e n tä r e v a n d e s ta n d v a n h e t o n d e r-
z o e k m e t b ijz o n d e r e a a n d a c h t v o o r maar. Nijmegen
Halliday, M.A.K.
1985 A n I n tr o d u c tio n to F u n c t i o n a l G r a m m a r . Amsterdam: North Holland
Hellwig, A.
1974 ‘Zur Funktion und Bedeutung der griechischen Partikeln’. G i o i t a 52, 145-
71
de Jong, I.J.F.
1991 N a r r a t i v e i n D r a m a . T h e A r t o f th e E u r i p i d e a n M e s s e n g e r - S p e e c h .
Leiden/New York: Brill
Kroon, C.H.M.

45 My thanks are due to Professor A.M. van Erp Taalman Kip, Dr P. Hatlie, Professor S.L. Radt, Dr
R.F. Regluit and Dr A. Rijksbaron for their comments on an earlier version.
 EMPHASIS AND AFFIRMATION 231

1992 ‘Particula perplexa’. In: G. Bakkum, C.H.M. Kroon, R. Risselada (eds),

Amsterdam: UvA, 53-64
P e n te c o s ta lia .
1994 ‘Discourse Connectives and Discourse Type: the Case of Latin a t '. In: J.
Herman (ed.), L i n g u i s t i c S tu d ie s o n L a t i n . (Studies in Language Companion
Series 28). Amsterdam/Philadelphia: Benjamins, 303-17
1995 D i s c o u r s e P a r t i c l e s i n L a t i n : A S t u d y o f nam, enim, autem, vero a n d at.
Amsterdam: Gieben
Levinson, S.C.
1983 P r a g m a tic s . Cambridge: Cambridge University Press
Rijksbaron, A.
1991 G r a m m a tic a l O b s e r v a tio n s o n E u r ip id e s ’ B a c c h a e . Amsterdam: Gieben
Ruijgh, C.J.
1971 A u t o u r d e ‘τ ε é p i q u e ’. E tu d e s s u r la s y n t a x e g r e c q u e . Amsterdam: Hakkert
Schiffrin, D.
1987 D i s c o u r s e M a r k e r s (Studies in Interactional Sociolinguistics 5). Cambridge:
CUP
Sicking, C.M.J.
1986 ‘Griekse partikels: definitie en ciassificatie’. L a m p a s 19, 125-41
1994 ‘Taalkundige observaties bij het onderwijs in het Grieks’. L a m p a s 27,4-27
Sicking C.M.J. & J.M. van Ophuijsen
1993 T w o S tu d ie s i n A t t i c P a r tic le U s a g e . L y s i a s a n d P la to . Leiden: Brill
Slings, S.R.
1980 “‘KAI adversativum”—Some Thoughts on the Semantics of Coordina­
tion’. In: D.J. V. Alkemadeet al., L i n g u i s t i c S tu d ie s O ffe r e d t o B . S ie r ts e m a .
Amsterdam, 101-25
Wakker, G.C.
1994 C o n d i t i o n s a n d C o n d itio n a ls . A n I n v e s t i g a t i o n o f A n c i e n t G r e e k . Amster­
dam: Gieben
1994a ‘μάν, μήν.’ In: L e x i k o n d e s f r ü h g r i e c h i s c h e n E p o s 15, 29-30
1995 ‘“Welaan dan dus nu”. Partikels in Sophocles’. L a m p a s 28, 250-70
1996 ‘The Discourse Function of Particles. Some Observations on the Use of
μάν/μήν in Theocritus’. In: M.A. Harder, R.F. Regtuit, G.C. Wakker (eds),
H e lle n is tic a G r o n i n g a n a II. T h e o c r itu s . Groningen: Forsten, 247-63
Weydt, H. (ed.)
1969 A b tö n u n g s p a r tik e ln . Berlin

Weydt H. & H. Ehlers

1987 P a r tik e l- B ib lio g r a p h ie . Frankfurt am Main: Lang
Wolski, W.
1986 P a r t i k e l l e x i k o g r a p h i e . E i n B e i t r a g z u r p r a k t i s c h e n L e x ik o lo g ie
(Lexicographica. Series 14). Tübingen: Niemeyer
 PARAPLEROMATIC LUCUBRATIONS

INEKE SLUITER
Vrijc Universitcit Amsterdam
Institute for Advanced Study, Princeton

Not unlike their modem counterparts, ancient linguists like neat and all-encompass­
ing classifications. Their main theoretical framework, the system of the parts of
speech, is designed to accommodate every instance of each and every word in the
process called μερισμός (‘parsing’). The p a r t e s o r a t i o n i s were defined by a combi­
nation of formal and semantic considerations, to which essential syntactic information
could be added (as in ‘adverbs tend to go with verbs’). The latter type of information
was considered part of the general semantics of a word class and tended to be con­
fined to observations on the physical combinations of words and their equally physi­
cal positions v i s - à - v i s each other. The main p a r t e s were subdivided into numerous
subcategories, again mostly on the strength of semantic criteria. Dionysius Thrax, for
instance, enumerates 26 different types of adverbs, all of which are purely defined by
their meaning, with the likely exception of the έπιρρήματα μεσότητος. It is probable
that the primary trait this latter group has in common is a morphological one, namely
their ending in -ως. Similarly, he distinguishes eight (or nine) types of
‘conjunctions’,1 all of which are defined by their function or their semantic load—
often the two are hard to distinguish (D. Th. 87. Iff.).
As with any system descriptive of language, the actual empirical material turned
out to be recalcitrant, and to resist complete pigeonholing in terms of the p a r t e s that
were distinguished. This appears from the elaborate discussions Apollonius Dysco­
lus devotes to the classification of problematic words, and which he habitually inserts
between his discussion of the definition, syntax and semantics of each part of speech
and the more detailed discussion of the morphological characteristics of specimina
belonging in each part. Such discussions are extant e.g. in his D e p r o n o m i n i b u s
( p r o n . 26.23 - 35.5) and D e a d v e r b i i s (a d v . 126.24 - 145.25), while a similar section
from D e c o n iu n c tio n ib u s ( c o n i.) got lost in the lacuna after coni. 214.26.12
However, the framework of the p a r t e s o r a t i o n i s itself offered some room to
accommodate borderline cases, in that it contained several categories which were
capable of absorbing precisely those words whose meristic characteristics were less
clear-cut. The adverb is probably the best example. Any word which is used in a way
that sets it off from the part of speech to which it would usually be taken to belong.
1 The ancient word-class of the σύνδεσμ οι encompasses both more and less than our term
‘conjunctions'. I will use the word in inverted commas to draw attention to this fact.
2 It is announced coni. 213.18 προσέτι και τά δόξαν έσχηκότα συνδέσμων, ού μήν δντα.
234 INEKE SLUITER

becomes an adverb.3 For instance, any declinable word which is used in an adverbial
way thereby turns into an adverb itself (a d v . 120. Iff.). As we will see, the subclassifi­
cation of the ‘conjunctions’ (σύνδεσμοι) contains a potential receptacle for residual
cases as well, the sub-set of the σύνδεσμοι παραπληρωματικοί.4
In the following I will argue that the theoretical distinction of this group of parti­
cles in antiquity develops out of the debate on whether or not ‘conjunctions’ have
meaning, and an interest in the rhetorical and stylistic effects of redundancy. Later
grammatical theory adopted the group with traces of its background in stylistic theory
shining through, and exploited its potential to function as a port-manteau category in
its word-class system.

‘Conjunctions’ (σύνδεσμοι) were a recognized linguistic category from Aristotle

onwards ( P o e t . 1456b39ff.), but the subcategory of the παραπληρωματικοί was a
later addition to linguistic theory. Summarizing the relatively late l o c u s c l a s s i c u s on
the topic (Ap. Dysc. c o n i. 247.22 - 258.26), the group can provisionally be defined
as consisting of particles with a wide range of meanings, but sharing the characteristic
that they are also (and even predominantly) used without any distinguishable seman­
tic or syntactic impact or purpose, to ‘fill out’ or embellish metre and style. Their
name derives from this common element, because it is more practicable than multiply­
ing the number of types in an inordinate way by naming every one of these
‘conjunctions’ after their very diverse meanings.
It is likely that distinction of this category was facilitated by more general dis­
cussions about the question whether or not s y n d e s m o i as a group had meaning.
Aristotle had categorically denied this,5 and traces of the debate can be found in Posi­
donius (a p u d Apollonius Dyscolus).6 Only after a consensus had been established
that the ‘conjunctions’ in general did have meaning, did the need arise for a separate
subcategory which could deal with those instances that to all intents and purposes did
not Yet, from the beginning stylistic implications are more important than semantico-
syntactic ones. [Ar.] P r o b i . XIX 20 (919a), often adduced in discussions of para-
pleromatic particles, is usually interpreted as an early recognition of the non-neces­
sary character of ‘some’ s y n d e s m o i , and this is taken to refer to their semantico-
syntactic qualities. However, a closer look at the context reveals that this is not its

3 The adverb was called pandectes for that very reason, e.g. Charisius 252.29ff. B.: cum adverbium
Sto ici... pandecten vocent, nam omnia in se capit quasi collata per saturam concessa sibi rerum
varia potestate·. Clemens Ars grammatica (ed. Tolkichn) 88.17ff.: Hoc quoque intuendum, quod
haec pars, id est adverbium, duo nomina habet apud Graecos, id est 'epirrhema', quod
interpretatur 'adverbium', et 'pandecien' quod interpretatur 'omne dictum' [this is wrong, of
course], quia omnis pars orationis cum desinit esse quod est, adverbium fit.
4 Hellwig (1974: 149f.) considers this category an ad hoc solution like the vague modem class of
‘particles'. Schenkeveld (1988) demonstrates that ‘particle’ is not an ancient concept.
5 Poet. 2 0 ,1456b38: σύνδεσμος δε έστιν φωνή άσημος, cf. 1457a4.
6 Coni. 214.4ff.: Ποσειδώνιος έν τφ περί συνδέσμων άντιλέγων πρός τούς φάσκοντας, ώς οί
σύνδεσμοι ού δηλούσι μέν τι, αύτύ δέ μόνον τήν φράσιν συνδέουσι [possibly the Stoic
‘orthodoxy’, cf. D. L. VII 58]... This is Posidonius the Stoic (135-50 B Q , not the grammarian, see
Baratin (1989: 25 n. 2); Atherton (1993: 305; 306 n. 77); for the passage, see Belli (1982); Sluiter
(1990: 117 n. 293).
 PARAPLEROM ATTO LUCUBRATIONS 235

primary intention, nor does it seem to envisage only those particles which were later
styled parapleromatic.
In this ‘Problem’, [Aristotle] is dealing with music and in particular with the
nature of the tone called m e s e . If this tone is out of tune, the whole melody will sound
unpleasant, whereas other tones would just sound unpleasant themselves under the
circumstances, but would not affect the whole. The author explains this as follows:

πάντα γάρ τά χρηστά μέλη πολλάκις τη μέση χρήται, καί πάντες οί

αγαθοί ποιηταί πυκνά προς την μέσην άπαντώσιν, καν άπέλθωσι, ταχύ
έπανέρχονται, προς δέ άλλην οϋτως οΰδεμίαν.
‘For all the best tunes make frequent use of the m e s e , and all good musicians
search it out frequently, and quickly revert to it, even if they leave it, but not to
any other note to the same extent’ (tr. Hett, adapted).

Then follows a linguistic analogy:

καθάπερ έκ των λόγων ένίων έξαιρεθέντων συνδέσμων ούκ εστιν ό

λόγος Ελληνικός, οΐον το τέ καί το καί, ένιοι δέ ούθέν λυποΰσι διά τό
τοίς μέν άναγκαΐον είναι χρήσθαι πολλάκις, εί έσται λόγος, τόΐς δέ μή,
οΰτω καί των φθόγγων ή μέση ώσπερ σύνδεσμός έστι, καί μάλιστα των
καλών, διά το πλειστάκις ένυπάρχειν τον φθόγγον αΰτης.
‘Just as when from discourse certain “conjunctions” are removed, the lan­
guage is not (good) Greek, e.g. t e and k a i , while others can be removed with­
out harm (for it is necessary to use some of them frequently for there to be
discourse at all, but for others that is not the case); in the same way the m e s e
is as it were the “conjunction” of musical sounds, especially of good music,
because its sound must have a very frequent place in it. ’

The author compares the m e s e to ‘conjunctions’ that need to be employed frequently,

and are opposed to ones that can be left out without harm. The distinction is not,
therefore, between α ν α γ κ α ίο ι and μή ά να γκα ίο ι as such, but between
‘conjunctions’ that need to be employed f r e q u e n t l y and those that do not. This is
made clear by the preceding statement about the m e s e which is used πολλάκις,
searched out πυκνά and reverted to ταχύ. The emphasis on style is confirmed by the
repeated reference to the quality of the music (πάντα τά χρηστά μέλη) and the com­
posers (πάντες ον αγαθοί ποιηταί). In the linguistic analogy, the author seems to be
alluding to patterns of expectancy created by certain ‘conjunctions’ (e.g. τ ε ... καί).
This would mean that not only parapleromatic particles, but also e.g. ‘conjunctions’
like γάρ belong in the group that can be removed without making the text incoherent.
When read this way, the text fits in perfectly with a whole series of early rhetori­
cal treatises imparting stylistic advice on the audience-friendly use o f s y n d e s m o i . 1

7 E.g. Anaximenes’ Ars Rhetorica 25 (1435a39ff.): μετά δέ συνδέσμους, ους δ ν προείπης, άπο-
δίδου τούς άκολουθοΰντας. τό μέν ουν συνδέσμους άποδιδόναι τούς άκολουθοΰντας τοιόνδε
236 INEKE SLUITER

Their correct use will lead to clarity and Hellenistnos. In Aristotle’s R h e t o r i c , the first
factor contributing to Hellenismos is said to consist in the right use o f σύνδεσ μοι
CR h e t. Ill 5, 1407al9ff.):

εσ τι δ' αρχή της λέξεω ς τό έ λ λ η ν ίζ ε ιν τούτο δ’ έσ τίν έ ν π έν τε, πρώτον

μ εν έ ν το ΐς σ υνδέσ μοις, α ν άποδιδω τις ώς πεφ ύ κασ ι π ρ ό τερ ο ι κ α ί
ύ σ τεροι γίγνεσ θ α ι άλλήλω ν, οιον εν ιο ι ά παιτοΰσ ιν, ώσπερ ό μ έν κ α ί ό
έγώ μ έν ά π α ιτεΐ τον δ έ κ α ί τόν ό δέ. δ ε ι δ έ εω ς μέμνηται άνταποδιδό-
ν α ι άλλή λοις, κ α ί μήτε μακράν άπαρτάν μ ήτε σύνδεσμον προ σ υν δέ­
σμου άποδιδόναι τού α ν α γκ α ίου - κτλ.
‘The beginning of good style is Hellenismos, pure Greek. This consists in five
elements, first in the use o f “conjunctions”, if one makes them correspond in
their natural order of priority and posteriority, the way some of them require.
E.g. m e n and e g o m e n require d e and h o d e . They should be m ade to corre­
spond to one another as long as they are still firesh in memory, and should not
be m ade to connect at a long distance, nor should one m ake a different
“conjunction” correspond instead of the necessary one.’

The desirable quick resolution o f the expectancy created by a syndesm os corre­

sponds to the way good composers quickly revert to the m e s e in the P r o b le m a ta .
The concept o f the paraplerom atic syndesmoi thus seems absent from the
pseudo-Aristotelian P r o b le m a ta , although the stylistic qualities of s y n d e s m o i in gen­
eral were recognized and would eventually be helpful in their distinction. (In fact, in
contrast to the position taken in the P r o b le m a ta , they would become the s y n d e s m o i
with a virtually exclusively stylistic impact p a r e x c e l l e n c e ). Neither can the πα ρ α­
πληρω μ ατικοί have commanded the particular interest o f the first Stoic language
theorists. Their theory of the ‘conjunction’ was lim ited to its logical uses for the
construction of complex propositions, which are precisely defined by the type of
‘conjunction’ connecting their parts (D. L. VII 71ff.). By their very nature, the para­
pleromatic ones would escape notice in that context. However, there are two contexts
that connect the Stoics with the σύνδεσμοι παραπληρω ματικοί, one vaguely, the
other explicitly. The first is in D. L. V II67.* 8 Here, in the Stoic version o f speech-act
theory, a distinction is made between a x io m a ta , the bearers o f truth and falsity, and
the form called ομοιον άξιώ ματι (‘similar to the axiom ’). The latter is defined as
follows:

έ σ τίν έγώ μ έ ν παρεγενόμην οΰ έφην, συ δ ε φάσκων ήξειν ούκ ήλθες- πάλιν όταν ό αυτός
σύνδεσμος άκόλουθος ή, οΐον- συ γάρ κώ κείνων αίτιος έγένου, καϊ τούτων α ίτιος σύ; further
Isocrates apud Syrianus in Hermog. 128.6 R.; Max. Planudes, V 469 Walz and Joannes Siccliota, VI
156.19 W. = Radermacher Artium Scriptores B XXIV 22; Spengcl, Συναγωγή τεχνών 154-72
believes this to be part of Isocrates’ τέχνη Ρητορική; contra Blass, Att. Beredsamk. I I 105; cf. G.
Mathieu (Budé edition IV, 233f.).
8 Professor Schenkevctd pointed out the relevance of this text to me.
 PARAPLEROMATIC LUCUBRATIONS 237

δ την εκφοράν έχον αξιωματικήν παρά τίνος μορίου πλεονασμόν ή

πάθος έξω πίπτει τοΰ γένους των αξιωμάτων, οιον

καλός γ' ό Παρθένων

ώς Πριαμίδησιν έμφερής ό βουκόλος.

‘that which having the form of an axiom falls outside the class of axioms
because it exceeds it by an extra word or by emotion, e.g.

Beautiful indeed the Parthenon!

How does the cowherd resemble Priam’s sons!’9

Note that neither the term σύνδεσμος nor that of παραπληρωματικός is used in this
context. Γε certainly qualifies as a parapleromatic ‘conjunction’ in later theory, but
ώς is an adverb. Note also that the examples are both poetic. 1 will return to this pas­
sage later.
The second instance stems from a period in which a more general overview of
grammar was extrapolated from Stoic work on logic, and attempts were being made
to provide language descriptions with a claim to exhaustiveness. In that context, we
know that the Stoic Chaeremon (a teacher of Nero’s), who wrote on ‘conjunctions’,
devoted some attention to the question of the classification of the parapleromatic
‘conjunctions’ in view of their alleged lack of meaning.10 It would seem that by this
period ‘conjunctions’ as such were agreed to have meaning—the subcategories in
Dionysius Thrax are after all semantic or functional in nature. It was the subcategory
of the parapleromatic ‘conjunctions’ that had become the focus for discussion of the
problematic notion of absence of meaning. This aspect was then combined with an
element inherited from earlier Peripatetic observations on the class of ‘conjunctions’
as a whole, namely an interest in the rhetorical and stylistic function of these syn-
desmoi.
The earliest attestations of the use of παραπληρωματικοί date from the 1st cent.
BC. We know that both Tyrannio (early 1st cent. BC) and Trypho (contemporary of
Augustus) discussed this class. P. Yale 1.25 (1st cent. AD) lists it as one of the
classes of the σύνδεσμοι. The discussions by Tyrannio and Trypho suggest that
Dionysius Thrax indeed knew this sub-category, although the part of his. T e d i n e that
contains its description amply postdates him. His description does not ascribe any
meaning to this category, but neither does it define them by the absence of meaning.
Rather, it describes them in functional terms as being used μέτρου ή κόσμου ένεκεν
‘for the sake of metre or ornament’ (D. Th. 96.3f.). Many ancient interpreters con­
nect this view on the παραπληρωματικοί with the clause in Dionysius’ over-all
definition of σύνδεσμοι (D. Th. 86.3f.):

9 Cf. Schenkeveld (1984: 303,307,315).

10 Chaeremon: Ap. Dysc. coni. 248.1.
238 INEKE SLUITER

(Σύνδεσμός έστι λέξις συνδέουσα διάνοιαν μετά τάξεως) καί τό της

ερμηνείας κεχηνός δηλοΰσα. (ν.1. πληρούσα)
‘(A “conjunction” is a word that connects the thought while expressing
order,) and it indicates (or: fills) the hiatus of the language’

They take it that τό κεχηνός refers to the collision of two vowels, which is judged to
be a stylistically unpleasant effect, and that the παραπληρωματικοί are the sub­
group that performs this syndesmic function p a r e x c e lle n c e .
The alternative and more sophisticated (but also more far-fetched) ancient inter­
pretation is that the last clause of the definition allows the inclusion of the διαζευ­
κτικοί or disjunctive ‘conjunctions’. These s y n d e s m o i ‘conjunct’ on the formal
level, but their meaning is to exclude, to separate off. It can however be described as
‘filling/indicating the gaps of discourse’, in the sense that they make the connection
of thought explicit.11 Baratin (1989: 38) generalizes the applicability of the clause
καί τό της ερμηνείας κεχηνός δηλοΰσα, interpreted along these lines, to all types
of conjunctions.1112 I find his reading convincing as an attempt at a charitable and
maximizing interpretation, but will concentrate in what follows on the ancient views,
which predominantly take the more obvious route and understand the clause as refer­
ring to hiatus.
The use of parapleromatic particles smoothens and remedies the harshness cre­
ated by hiatus, and renders the language more euphonic. Of the technical grammari­
ans, notably Trypho is of this opinion, and he famously compared parapleromatic
‘conjunctions’ with ‘padding (στοιβή) to prevent jarring and breaking of
amphoras’. 13 When Apollonius Dyscolus is describing this euphonic use of
‘conjunctions’, he compares it to that of the ephelcustic -v-. Interestingly, the use of
such a -v- is referred to in p r o n . 50.11 as:

σαφές δτι τό χ α σ μ ώ δ ε ς των φωνηέντων ά ν α π λ η ρ ώ ν τη τού ν προσθέσει

(sc. Homerus)
‘clearly filling out the hiatus of the vowels by the addition of the n u \

11 See Uhlig ad D. Th. 86.3-4, and e.g. Sch. D. Th. 103.9-12; 436.23.
12 He seems unaware of the anticipation of his interpretation by the Scholiast on D. Th. 436.30ff.
13 Ap. Dysc. coni. 252.31; 253.2; 253.9f. In 252.32 Apollonius is clearly alluding to Trypho's
description of these ‘conjunctions’ as ‘padding’, showing that he derives his own views on their
euphonic use from him (although he rejects the view that a whole part of speech should owe its
existence to none other than euphonic reasons). Cf. 247.26ff„ esp. 247.29 (Trypho) ύπέρ τού τα της
φράσεως μή τραχύνεσθαι. Apollonius claims that Trypho added a clause to his definition of the
σύνδεσμοι as a group in order to be able to include these meaningless particles, which do not
strictly speaking do any ‘connecting’. The clause he mentions is και τό κεχηνός της έρμηνείας
έστιν όπου παραπληρώυ, coni. 247.24Γ, alluded to in synt. 378.5f. on the parapleromatic
conjunctions: άναπληροΰν tò κεχηνός της έρμηνείας. Notice the resemblance with the Dionysian
formulation. Baratin 1989: 35 has claimed that the form of the clause (with the masculine participle
παραπληρών) makes it unlikely that it actually belonged in the definition of the whole pars
oralionis, since in that case one would have expected either a feminine form to correspond with
λέξις, or a neuter to correspond with μέρος λόγου. This is very ingenious, but one cannot exclude
the possibility that Apollonius' quotation has undergone ‘grammaticalization of the lemma’, which
would typically yield the nominative masculine (cf. Tosi 1988).
 PARAPLEROMATIC LUCUBRATIONS 239

thus illustrating how naturally this interpretation fits in with normal linguistic usage.
Indeed, there can be no doubt that the natural reading of τό χασμώδες and το κεχη-
νός is that it refers to hiatus between vowels.14
The avoidance of hiatus takes us straight back to stylistics. This fits right in with
the fact that some of our earliest testimonies for the παραπληρωματικοί (as
opposed, this time, to the group of s y n d e s m o i as a whole) also take the form of
stylistic recommendations. Protesting against their idle use, pseudo-Demetrius pre­
scribes them only if one wants to produce an effect of grandeur { E lo c . 55ff.):

Τόίς δε παραπληρωματικϋΐς συνδέσμοις χρηστέον ούχ ώς προσθήκαις

κενάις καί οιον προσφύμασιν ή παραξύσμασιν, ώσπερ τινες τω δή
χρώνται προς ούδέν καί τω νυ καί τω πρότερον, άλλ' αν συμβάλλωνταί
τι τω μεγεθει τοΰ λόγου ...
O ne should use the parapleromatic “conjunctions” not as empty additions,
and as it were as abnormal growths or superfluous polishing, the way d é and
n u and p r o t e r o n are sometimes used without a purpose, but only if they con­
tribute something to the grandeur of the language’

One way of doing this is to use δή to amplify the beginning of a statement—this

emphasizes the divide between two statements and makes for a dignified impression
{ E lo c . 56). Another function of the particle δή is to invest discourse with p a t h o s
{ E lo c . 57). Pseudo-Demetrius refers to Praxiphanes, a student of Theophrastus’, who
is scornful of those who προς ούδέν άναπληροΰντες ... τον σύνδεσμον (‘use the
“conjunction” as an idle filler’) { E lo c . 58). This is another indication of the stylistic
interest taken in ‘conjunctions’ in general in the Peripatos, but although the context is
suggestive, it cannot be proven that Praxiphanes was thinking of the parapleromatic
‘conjunctions’, identified as a separate group. Praxiphanes is not pseudo-Demetrius.
He compares the idle use of words like νυ, but also like πρότερον, with actors’ inter­
polations of interjections like φευ and α'ί d i used e x tr a m e tr u m . (In fact, such a bad
use of these s y n d e s m o i would probably make them fall outside the scope of Diony­
sius Thrax’ definition, since they do not fill out the metre at all, but ruin it in a way.)
In general, the whole context—as far as it can be made out from the poor transmis­
sion—emphasizes not the mere fact of άναπληροΰν, or of being a προσθήκη, but the
use of προσθηκαι κ ε ν α ί { E lo c . 55), προς ούδέν (58) or μάτην (58). Such a usage
violates the rhetorical requirement of τό πρέπον, and this is confirmed by the use of
έπρεψεν in 57 for an acceptable use of the same particle.
The text in pseudo-Demetrius rewards further analysis. The terms πρόσφυμα
and παράξυσμα have no parallels in a stylistic context. Both are used metaphorically,
much like στοιβή ‘padding’ was in Trypho, a πρόσφυμα being any kind of growth

14 Cf. e.g. Sch. D. Th.. 146.31 ff.: καί έκθλιψις μέν έστιν, ήνίκα εύρεθη λέξις ε ις φωνήεν η
φωνήεντα καταλήγουσα, της έπιφερομένης λέξεω ς άπό φωνήεντος ή άπό φωνηέντων
άρχομένης· τότε γάρ < διά> τό χασμώ δες καί κεχηνωδές έκθλίβεται τό τέλος της προ­
ηγούμενης λέξεως. Clearly, we are dealing with well-established technical terminology here.
240 INEKE SLUITER

that is ‘in addition to (the natural constitution?)’ (πρός),15 while παράξυσμα may
either mean ‘extra scratches, extra letters’,16 or maybe ‘a result of excessive polish­
ing’.17 Παρά- in παράξυσμα points at the undue, excessive or redundant nature of
the polishing. All three terms, προσθήκη, πρόσφυμα and παράξυσμα underline by
their pseudo-etymologizing the essential redundancy of the παρα-πληρωματικός.
Whereas compounds of πληρόω are routinely used to denote syntactic or semantic
completion, i.e. the necessary completion of a semantic and syntactic structure, a
παραπληρωματικός does its ‘filling out’ over and above what is strictly speaking
required or necessary in those respects.
Unlike πρόσφυμα and παράξυσμα, προσθήκη resonates widely in the field of
stylistics. Dionysius of Halicarnassus describes a certain κώλον in Plato as an
unfortunate προσθήκη, since it is neither necessary (άναγκαία) nor used ‘for the
sake of beauty, or another form of additional ornamentation. ’18 These possibilities
are resumed a little later as μήτε του αναγκαίου χάριν ... μήτε του περιττού.19
These are the only two acceptable forms of προσθήκη. Otherwise, an addition turns
into άκαιρία, ‘bad timing, bad taste’. Clearly, the word ‘redundant’ (περιττός) by
itself does not exclude a contribution to beautification, as did the pseudo-Demetrian
qualification πρός ούδέν.
Further, προσθήκαι are a stock-issue in the context of the theory of the three
styles. Firstly, the sublime or severe style (αύστηρά αρμονία) is not served by addi­
tions that smoothen stylistic ruggedness, since that is the very core characteristic of
this style. Its aim is not to be periodical, so it avoids προσθήκαι that do not con­
tribute to the sense.20 Interestingly, the same concept is also expressed using παρα­
πλήρωμα (D. H. D e m . 39.212.20ff.):

(the severe style uses asymmetrical periods), μηδέ γε παραπληρώμασι των

ονομάτων οΰκ άναγκαίοις ώς πρός την ύποκειμένην διάνοιαν χρωμενας
‘without using filler words that are not necessary for the underlying meaning’

15 One is strongly reminded of Sch. D. Th. 356.15ff., explaining why the Stoics had not recognized
the adverb as a separate part of speech: τα γάρ Επιρρήματα o w e λόγου ούτε αριθμού ήξίωσαν,
παραφυάδι καί έπιφυλλίδι αύτά παρεικάσαντες. Έπιφυλλίς refers to the small grapes left for
gleaners: it was used metaphorically for bad poets in Ar. Ran. 92 (cf. n. 17), quoted in D. H. Ars
Rhet. X 18 to describe epilogues in a deprecating way after first having called them a sort of dessert
after the main meal. The compound with έπ ι-may again have been found attractive for a description
of Επιρρήματα, even though it is precisely denied that the Stoics recognized those words as a
separate part of speech. Παραφυάς ‘side-growth’ is an interesting parallel for πρόσφυμα, the more
so since at least one of the examples cited by Demetrius decidedly looks like an adverb (πρότερον),
see below.
16 Ξύσματα = γράμματα, Hsch.
17 Rather than the ‘filings’ or ‘shavings’. Cf. for the metaphor Cic. Brut. 93; Hör. Ars 291; Ον. Tr.
1.7.30 (ultima lima)·. Quint. 10.4.4. An interesting verbal parallel is Ar. Ran. 881 (cf. n. 15)
παραπρίσματά τ’ έπών ( ‘sawdust, sawn-off bits’, Dover a.I.), where the παρα- element seems less
emphatic than in παράξυσμα.
18 Κ άλλους... ή τών άλλων τινός τών έπιθέτων ένεκα κόσμων, Dem. 24.182.5ff.
19D. Η. Dem. 24.182.20f.
20 D. H. CV 22.97.12: (αύστηρά άρμο νία)... ούτε προσθήκαις τισιν ονομάτων, ινα ό κύκλος
έΚ7ΐληρωθη μηδέν ώφελούσαις τον νοΰν χρωμενη ούτε κτλ.
 PARAPLEROMATIC LUCUBRATIONS 241

The αυστηρά άρμονία is όλιγοσύνδεσμος and άναρθρος.21 And it even cultivates

hiatus.22 Grandeur (μέγεθος) is one of the intended effects of the grand style.
Longinus {D e s u b ì. 21.2) points out that ‘s y n d e s m o i and other προσθηκαι’ detract
from pathetic effects that sit well with this style—partly disagreeing with pseudo-
Demetrius {E lo c . 57, see above).
The Stoic speech act ομοιον άξιώματι, defined by the presence of a redundant
word or of πάθος and illustrated by poetic examples (see above p. 236), hovers at the
intersection of logic and the type of stylistic observations discussed here. The poetic
examples indicate that constructions that are directly relevant to logic are being sup­
plemented (in order to arrive at something like an exhaustive description) by material
drawn from the common stock of the student of language and literature, namely
poetry. An emphasis on the presence of redundancy v i s - à - v i s the regular axiom in
this specific type is confirmed by Sextus Empiricus’ description of the same type of
sentence as πλέον τι άξιώματος {Μ . 8.73). In both cases, however, the formal obser­
vation is made subservient to a general focus on logic: the speech act is defined with
direct reference to the αξίωμα pure and simple, the bearer of truth and falsity and the
central speech-act type in Stoic logic.23 A prudent conclusion would be that the
Stoics’ wide interpretation of the scope of their dialectic led them to make observa­
tions on form and style over and above the requirements of logic, thus providing
stimuli for later students of linguistic phenomena, who emphasized the stylistic side
and downplayed that of logic.
To return to προσθηκαι and stylistic theory, in the elegant style (άρμονία
γλαφυρά) προσθηκαι are useful and necessary to achieve the full intended stylistic
effect. This style strives after ήδονή and κάλλος rather than rugged grandeur, and
προσθηκαι that are not strictly speaking necessary for the sense contribute to that
purpose.24 Again, προσθήκη may be replaced with its synonym παραπλήρωμα.25
Secondly, a pleasant effect is contrived by euphony,26 and again προσθηκαι are a
helpful tool.27 And thirdly, the άρμονία γλαφυρά avoids hiatus by adding words that
again are not strictly speaking necessary, but prevent roughness (D. H. D e n t .
40.215.19ff):

21 D. H. CK22.98.lf.; cf. Dem. 39.213.6ff.

22 D. H.Dem. 38.210.14ff.
23 Albert Rijksbaron pointed out to me that in the sentence ‘how like Priam's sons the cowherd is*
the truth of the underlying axioma is presupposed—in that sense the sentence truly conveys ‘more
than an axioma’. This is true and relevant. However, it only goes for die examples with ώς, not for
those with ye.
24 D. H. CV 9.33.?ff.: άλλα μην άτι γ ε καί μετασκευάς δέχεται τών κώλων ένια τοτέ μέν
προσ&ήκας λαμβάνοντα ούκ αναγκαίας ώς προς τον νούν, τοτέ δε αφαιρέσεις ατελή ποιού-
σας τήν διάνοιαν, ας ούκ άλλου τίνος ένεκα ποιοϋσι ποιηταί τε κα ί συγγραφείς ή της
αρμονίας, ίν ' ήδεία καί καλή γένηται, πάνυ όλίγου δείν οΐομαι λόγου.
26 D. Η. CV 9.33.23f.: τίς ούκ άν φαίη παραπληρώματι λέξεως ούκ άναγκαίφ προσηρανίσθαι;
(sc. a period from Plato).
26 D. H. Dem. 40.214.23.
27 CV 102f., 67.5ff.: παραπληρώμασιν εύφώνοις.
242 INEKE SLUITER

και δήτα και παρεμβάλλειν αύ τάίς <άναγκαίαις> τινάς έτέρας λέξεις

υπομένει πρός ιό ν υποκείμενον νουν οϋτ' αναγκαίας ούτ’ ίσως
χρησίμας, δεσμού δέ τίνος ή κόλλης τάξιν τάίς πρό αυτών καί μετ’ αυτός
κειμέναις όνομασίαις παρεξομένας, ϊνα μή συναπτόμεναι προς άλλήλας
α ί καταλήγουσαί τε εις τραχύ γράμμα καί αί την αρχήν άπό τίνος
τοιούτου λαμβάνουσαι σπαδονισμούς των ήχων ποιώσι καί άντιτυπίας,
τή δε παρεμπιπτούση λέξει προσαναπαυόμεναι μαλακούς φαίνεσθαι
ποιώσι τούς ήχους καί συνεχείς.
‘Indeed, [the smooth style] is quite prepared to allow unnecessary words to be
added to the necessary which contribute nothing to the underlying sense, and
perhaps have no useful purpose, but are intended to serve as a sort of connec­
tion or bonding between what precedes and what follows, so that words ending
and words beginning with rough letters may not clash, choking the sound and
producing dissonance. The intervening phrase provides a rest and makes the
sound appear soft and unbroken’, tr. Usher

Dionysius proceeds to state that in this style the aim is ‘to draw and weave together
(συσπασθηναι, συνυφάνθαι) all the members of the period, achieving the impres­
sion of one continuous sound-stream (μιας λέξεως)’ (ibid. 216.8ff.).28 The
description sounds as a paraphrase of the concept of the σύνδεσμος παραπληρω­
ματικός: παρεμβάλλειν and παρεμπιπτούση; the fact that they do not make a nec­
essary contribution to the meaning; and the function of the intercalated words as
δεσμός and κόλλα. The description of how roughness is avoided is also reminiscent
of Trypho’s view of these words as ‘padding’.29 However, there is no compelling
evidence that Dionysius has the restricted group of the παραπληρωματικοί in mind,
and rather a lot of circumstantial evidence to suggest that he does not.30

28 There is a faint echo of Aristotle’s description of the λέξις ειρομένη as συνδέσμω μία (Rhei.
III 9 , 1409a24f.; cf. the frequent description of the Iliad as a λόγος that is συνδέσμω εις, e.g. Met.
B 4 , 1030bl0; H 6 , 1045al3; Poet. 2 0 ,1457a29), but in the case of D. H. we arc dealing with an
effect of sound rather than structure, and as such it is incorporated in a description of the periodic,
not the concatenated, style.
29 Cf. also τραχύ (216.3); σπαδονισμούς, άντιτυπίας, (216.4Γ).
30 Schenkeveld (1983: 71f.) lists the uses of σύνδεσμος in D. H. Although D. H. is familiar with
the term, he uses it in a rather loose way; thus it can include prepositions like έ π ί and έν
(Schenkeveld 1983: 74). Schenkevcld docs not discuss D.H.’s use of sub-groups of ‘conjunctions’,
because there is none. The combination σύνδεσμος παραπληρωματικός does not occur. In fact, all
his uses o f παραπλήρωμα and προσθήκη suggest that they express a more general stylistic phe­
nomenon, and arc not connected with a specific type of ‘conjunction’ as a technical term. The more
general usage o f δεσμός and κόλλα in Aristotle also points in this direction. Δεσμός and κόλλα are
two of the means by which unity is achieved, e.g. Ar. Met. Z 2, 1042bl6ff. (συνθέσει, κράσει,
δεσμφ, κόλλη, γόμφφ); Met. I 1, 1052a24 όσα κόλλη ή γόμφφ ή συνδέσμφ. Interestingly, these
metaphors are picked up and applied to all the lesser parts of speech by Ammonius In Ar. Int., CAG
IV 5.12.25ff.: ώσπερ γάρ τής νεώς α ί μεν σανίδες είσ'ι τά κυρίως μέρη, γόμφοι δέ καί λίνον
κα ί πίττα συνδέσεως αυτών καί της τού όλου ένώσεως ένεκα παραλαμβάνονται, τόν αυτόν
τρόπον κάν τφ λόγψ σύνδεσμοι καί άρθρα καί προθέσεις καί αυτά τά επιρρήματα γόμφων
τινών χρείαν άποπληρούσι; cf. I3.3ff. Obviously, in none of these cases does the application of
the metaphor envisage the parapleromatics.
 PARAPLEROMATIC LUCUBRATIONS 243

Undue application o f these stylistic devices to produce a fully periodic style at all
cost and irrespective of whether the subject-matter lends itself to it can lead to criti­
cism, which Isocrates in particular comes in for (D. H. I s o c . 3.58.18ff.):

ού γάρ άπαντα δ έχετ α ι ούτε μήκος τό αύτό ούτε σ χήμα το παραπλήσιον

ού τε ρυθμόν τόν ίσ ον, ώστε α ν ά γκ η πα ραπληρώ μ ασ ι λ έξεω ν οΰδέν
ώφελουσών χρ ή σ θ α ι κ α ί άπομηκύνειν πέρ α τού χρησίμου τον λόγον.
‘For not everything lends itself to the same length o r a sim ilar form or the
same rhythm. And so it becomes necessary to use fillers o f words that do not
have any use, and to draw out the speech over and above what is useful’

Now, if we compare the texts adduced so far, it becomes clear that the παραπληρω­
μ ατικοί σύνδεσμοι were associated particularly closely with a stylistic function that
could also be fulfilled by other parts of speech or even whole phrases: if applied well,
if fills out the sentence, not by the necessary completion of the thought, but by mak­
ing it smoother, remedying hiatus, and providing supple transitions. Thus, it is the
sub-group of the παραπληρω μ ατικοί, rather than the σ ύνδεσ μοι as a whole, that
becomes the focus for remarks on absence of meaning.
Our texts make it equally clear that the delimitation of the group must have been
vague originally, when παραπλήρωμα was a general stylistic qualification for a word
(group) that was strictly speaking redundant .31 Of the ‘literary critics’, only pseudo-
Demetrius uses the t e r m i n u s te c h n ic u s παραπληρωματικός σύνδεσμος. And in fact,
corrupt transmission o f pseudo-Demetrius was promptly assumed to get rid o f the
example πρότερον. But again, a comparison with the other stylistic criticisms seems
to suggest differently: when Longinus points out that it is hard to convey πά θος
when the style is being made smoother by συνδέσμων κ α ί των άλλω ν προσθηκών,
the sentence he quotes as an example contains words like κ α ί μήν, and γ ε μην, but
also πρώτον μέν, ειτα δέ, and another ειτα (21.1). In Lucian’s harangue against
Atticism, L e x ip h a n e s , the main character is given an emetic to get rid of his redundant
Atticistic phrases, which include μών, κφτα, δήπουθεν and αττα .32 While later the­
ory does not call these words σύνδεσμοι παραπληρωματικοί, it will point out that
they are redundant. Comm on terms are π α ρ έ λ κ ε ιν , π λ ε ο ν ά ζ ε ιν , α ρ γό ς, έ κ
π ερ ιττ ο ύ , περιττός, π ερ ιττ ε ύ ε ιν , κενός, κ ενώ ς π ρ ο σ κ έίσ θ α ι ,33 έ κ πλήθους .34

31 In Sch. D. Th. 461.15ff. παραπλήρωμα is one in a series of slylistic/rhetorical phenomena, and it

is described as: φράσις ή λέξις έκ περισσού λαμβανομένη, ώς τό <Β 493> άρχούς αύ νηών
έρέω νηάς τε προπάσας· ή γάρ πρό παρέλκει. Note that preposition and prepositional prefixes
were considered σύνδεσμοι προθετικοί by the Stoics. Traces of this usage can be seen in D. H„ cf.
previous note.
32 Μών, κ$τα: both are σύνδεσμοι άπορρηματικοί in D. Th. (94.2f.); Ap. Dysc. coni. 229.19ff.
(μών = διαπορητικός and contains οΰν παραπληρωματικός: κ$τα= used instead of καί or it is
διαπορητικός, although there is some discussion about its status. It could also be an adverb, since
it contains ειτα (so Trypho)).
33 Hipp. In Aratum 156.4; 8f. (ed. Manitius, BT) - προσαναπλήρωμα.
34 Cf. Ruijgh (1971: § 60); Ruijgh points out that παραπληρωματικός is not synonymous with
these terms, which always denote redundancy in a certain context, i.e. as an accidental feature, while
παραπληρωματικός is used as a classificatory term (1971: § 65). The two exceptions where τε is
 INEKE SLUITER

Elink Sterk has collected all the words whose redundant use has been noted by
ancient Greek scholiasts, glossographers and Atticists (II 41ff.). Among the exam­
ples we find εΐτα, έπεί, έπειτα, δήπου, δήθεν, as well as numerous other σύν­
δεσμοι and adverbs, but also a word like αττα, singled out by Lucian.35 Again, while
there may often be critical overtones in designating the use of a word ‘redundant’, the
terms as such do not exclude the possibility that their effect is to beautify the lan­
guage.36
Thus, there is a long-standing stylistic awareness of the incidental ornamental,
meaningless use of words that do signify in other contexts. If their ornamental effect
is doubted, such a use constitutes a stylistic f a u x - p a s . The part of speech in which
this phenomenon was observed to occur most frequently was the conjunction. Even
after it had been established that conjunctions in general are not meaningless (like
Aristotle had contended), but express the relationships—which have a reality of their
own—between states of affairs (Stoa),37 the discussion about meaningless words still
naturally gravitated towards this part of speech, but it concentrated on a sub-group,
the παραπληρωματικοί. Apollonius Dyscolus still feels the necessity of vigorously
combating a majority view that these words have no meaning at all.38 Supporters of
that idea may have felt backed up by the very name of the sub-group, but Apollonius
explains the name as a simple matter of classificatory convenience. There is no point
in endlessly increasing the number of subcategories, each covering a sub-set of pos­
sibly one word only. The system of grammar should be kept simple, elegant and easy
to memorize. These words have meaning, but they are rarely used for that. Usually,
their euphonic function explains their appearance, without their making any contribu­
tion to the meaning; even so they are useful, and this is the use they are named for.39
Apollonius’ parallel for this terminological solution of convenience is the word
υποτακτικός for ‘subjunctive’: this mode expresses a number of semantic values,
but the common formal characteristic of always following a σύνδεσμος determines
the name.40

called παραπληρωματικός, are Ap. Dysc. adv. 148.5; Sch. Horn. 0 124 (Ruijgh 1971: § 68f.). Note
that Elink Sterk is confused about the use of έκ πλήρους, which docs not mean that the word used
is redundant, but that it is spelled in scriptio plena, i.e. without apocope (e.g. II 48 γ ε , not γ'; 66
τε, not τ').
35 Ειτα: Sch. on Aristoph. PL 79, Elink Sterk I I 51; αττα: Harpocration etc., Elink Sterk II 44.
36 Cf. e.g. the use of περιττός signaled above (p. 240), in D. H. Dem. 24.182.20f.; cf. D. H. CV
9.33.12ft.: τις γάρ ούκ αν όμολογήσαι τηνδε μεν την λεξιν ήν ό Δημοσθένης είρηκε προσθήκη
πλεονάζειν ούκ άναγκαΐςι της άρμονίας ένεκα; cf. 33.21 χαριέστερον.
37 Frede (1977: 74), (1978: 62ff, 6 5 f), Atherton (1993: 302ff).
38 Ap. Dysc. coni. 247.22ff.
39 Coni. 252.29ff. It should be pointed out that Elink Sterk's otherwise useful contribution is fun­
damentally vitiated by his view that a word without meaning cannot have anything to contribute to
the beauty of a text (better Jahn 1847; 37ff.; Kroon 1995: 37 n. 6). Therefore, he refuses to allow
that the παραπληρωματικοί, even if just used qua παραπληρωματικοί (i.e. precisely in the cases
where the name is not a misnomer), can be meaningless.
40 Ap. Dysc. synt. 377.8ff; D. H. also knows this principle in naming the elements and the three
styles: they are never pure, but are named after the dominating characteristic, ώνόμασται δ' έκαστον
αυτών κατά to πλεονάζον, Dem. 37.209.21ff.
 PARAPLEROMATIC LUCUBRATIONS 245

I began this article by pointing out that the ancient meristic system allowed room for
residual cases: the class of adverbs formed a natural receptacle for them, and within
the class of the s y n d e s m o i , so did the parapleromatic ones. All s y n d e s m o i which
would otherwise have formed a semantic class of their own fitted into this category.
In this way it contributed to the economy of the system. The reason why paraplero­
matic uses of the s y n d e s m o s were not classified as adverbs is clear. Μερισμός is
primarily decided on the ground of meaning.41 If a word incidentally behaves as if it
belonged to a different word class, that instance will be classified accordingly. But
parapleromatic uses of the s y n d e s m o s do not have any distinguishable meaning at all,
and therefore retain their classification, as would happen with any other pleonastically
used part of speech. The difference is that with other words pleonastic use is the
exception, while here it is the rule, to the point where the whole sub-group is named
for this use. Chaeremon allows the formal behaviour of these words (i.e. the fact that
they are homonyms of regularly signifying ‘conjunctions’) to be the determining
factor in their μερισμός.42 Apollonius strengthens this argument by putting their
redundant use on a par with that of any other part of speech, and pointing out that
incidental redundancy is never a reason to assign a word to a different part of speech
( c o n i . 249.22ff.). He then proves that every σύνδεσμος παραπληρωματικός is also
used in a meaningful way ( c o n i. 249.31ff.), thus putting this sub-group firmly in the
realm of the normal and regular.
In this contribution I looked at the grammatical class of the σύνδεσμοι παρα­
πληρωματικοί from a specific angle. Two factors contributed to its development.
One is that discussions about whether or not conjunctions have meaning were
replaced with or focused on the question of whether or not this specific residual
group had any. Here I emphasized the economy of the grammatical system, which
has its own requirements that can take precedence over linguistic realities. The other
factor is the discussion about the stylistic impact of conjunctions. Here, too, we
observed a shift from early rhetorical admonitions about the most effective use of
corresponding s y n d e s m o i to an interest in an issue that directly affected the distinc­
tion of the parapleromatics: the rhetorical and stylistic interest in redundancy and its
effects. This was an old issue, but here too, the focus had to be narrowed. Virtually
any word or even phrase could be used redundantly, but it seemed especially relevant
in the case of the s y n d e s m o i , if no longer for the whole group, then certainly for this
sub-set that even got to derive its name from it.
Not only particles, but also articles can be parapleromatic. They are not neces­
sary, but hopefully they contribute to μέτρον, κόσμος or both of the book they are
helping to fill. Certainly there can be no άκαιρία in their expressing their admiration
for the author of the Odyssey around epic τε.

41 Sluiler (1990: 64). Ap. Dysc. pron. 67.6: ού γάρ φωναις μεμέρισται τά τοϋ λόγου μέρη,
σημαινομένοις δέ.
42 Apud Ap. Dysc. coni. 248. Iff.
 INEKE SLUITER

Bibliography

Atherton, C.
1993 T h e S t o ic s o n A m b ig u ity . Cambridge
Baratin, M.
1989 L a n a is s a n c e d e la s y n ta x e à R o m e . Paris
Belli, G.
1982 ‘Aristotele e Posidonio sul significato del “syndesmos”’. A e v u m 61, 105-7
Elink Sterk, J.W.
1845-46 ‘Grammatica Zetemata —De Parapleromaticis’. S y m b o la e L ite r a r ia e
7, 3-63 [=1]; 8, 3-71 [=11]
Frede, Μ.
1977 ‘The Origins of Traditional Grammar’. In: R.E. Butts and J. Hintikka (eds),
H i s t o r i c a l a n d P h i l o s o p h i c a l D i m e n s i o n s o f L o g ic , M e t h o d o l o g y a n d
P h i lo s o p h y o f S c ie n c e . Dordrecht, pp. 51-79
1978 ‘Principles of Stoic Grammar’. In: J.M. Rist (ed.). T h e S t o i c s . Berkeley-
Los Angeles-London, pp. 27-75
Hellwig, A.
1974 ‘Zur Funktion und Bedeutung der griechischen Panikein’, d o t t a 52,145-
71
Jahn, C.F.
1847 G r a m m a tic o r u m g r a e c o r u m d e c o n iu n c tio n ib u s d o c tr in a . Greifswald
Kroon, C.M.
1995 D i s c o u r s e P a r t i c l e s i n L a t i n . A S t u d y o f nam, enim, autem, vero a n d at.
Amsterdam
Ruijgh, C.J.,
1971 A u t o u r d e ‘τ ε é p i q u e ’. E tu d e s s u r la s y n t a x e g r e c q u e . Amsterdam
Schenkeveld, D.M.
1983 ‘Linguistic Theories in the Rhetorical Works of Dionysius of Halicarnas­
sus’. d o t t a 61, 67-94.
1984 ‘Studies in the History of Ancient Linguistics II. Stoic and Peripatetic
Kinds of Speech Act and the Distinction of Grammatical Moods’.
M n e m o s y n e 37, 291-353
1988 ‘From Particula to Particle - the Genesis of a Class of Words’. In: I. Rosier
(éd.), L 'h é r i t a g e d e s g r a m m a ir ie n s la t ì n s d e l ’a n tiq u ité a i a l u m iè r e s . A c te s
d u C o llo q u e d e C h a n tilly , 2 - 4 S e p te m b r e 1 9 8 7 . Paris, pp. 81-93
Sluiter, I.
1990 A n c i e n t G r a m m a r i n C o n te x t. C o n t r i b u t i o n s t o t h e S t u d y o f A n c i e n t L i n ­
g u is tic T h o u g h t. Amsterdam
Tosi, R.
1988 S tu d i s u lla t r a d iz io n e in d ir e tta d e i c l a s s ic i g r e c i. Bologna
 LES PARTICULES KE(N) ET A N DANS LES FORMULES DE
L’ÉPOPÉE HOMÉRIQUE

Paul W athelet
Université de Liège

Le grec a disposé d ’une particule κε, apparentée à l’élément qu’on retrouve dans
l ’adverbe κεί ou ε κ ε ί.*1 À l’origine, la particule κε avait le sens de a l o r s , à c e
m o m e n t f u t u r . Elle a connu un élargissement en -v, ce qui a donné κεν. Au degré
zéro, κεν prenait la forme * k n > κ α devant consonne et xdv devant voyelle.2 Il est
parfois arrivé que la séquence ού καν devienne, par mécoupure (et par un effet de
l’alternance ού/ούκ), ούκ αν,3 avec l’apparition d’une particule άν, qui a joué le
même rôle que κε(ν) dans d’autres dialectes.
Avec son sens premier de a l o r s , κε a servi à distinguer les subjonctifs qui
avaient valeur énonciative et qui étaient proches du futur, de ceux qui avaient un sens
directif. Les emplois du subjonctif avec κε(ν) étaient proches de ceux du futur de
l’indicatif en voie de création. Ainsi s’explique la présence dans la langue épique de
κε(ν) avec le futur de l’indicatif.
De la même manière, κε(ν) a été utilisé avec l'optatif. La présence de la particule
indiquait que l’optatif ne marquait ni un souhait, ni un regret, mais qu’il avait valeur
énonciative.
Je reviendrai plus bas sur l’évolution des emplois de κε(ν) et de αν.
Comme souvent les dialectes grecs ont privilégié l’une ou l’autre forme des par­
ticules κε, κεν, κά, αν. Le dorien au sens large emploie κά,4 l’éolien utilise κε(ν),5
de même que le cypriote.6 L’arcadien atteste άν, mais aussi κάν.7 L’ionien-attique
connaît άν. La langue épique utilise κε{ν) et âv, dans une proportion d’environ
quatre κε(ν) pour un άν.

Je liens à remercier ici mon collaborateur J.-M. Renaud, qui m'a apporté une aide décisive dans le
dépouillement et l’examen des formes.

1 C J. Ruijgh, ‘L’emploi le plus ancien et les emplois plus récents de la particule κε/σν’. Dans: La
langue et les textes en grec ancien. Actes du Colloque Pierre Chantraine (Grenoble 5-8 septembre
1989). Amsterdam: Gieben, 1992, p. 75-88.
2 Par application de la loi de Sicvers-Edgerton.
3 Sur l’étymologie de ού/ούκ/ούχ, cf. P. Chantraine, Diet, ét., p. 835, s.v. ού (cf. arménien o£).
4 Sur la question d elà longueur de l’a d e κα dans les parlers doriens, cf. M.T. Molinos Tejada, ‘La
particule modale KA dans la littérature dorienne’. REG 105 (1992), p. 328-348.
5 A. Thumb-A. Scherer, II, p. 109, § 257,6b.
6 A. Thumb-A. Scherer, II, p. 174, § 276,5b.
7 A. Thumb-A. Scherer, II, p. 140-141, § 2 6 6 ,5 c-e'ik âv : IG , V,II, 3 (E. Schwyzer, Del 3, 654), 1.
16,23, 31, etc. — Tégée, vers 400 av. J.-C.
 PAUL WATHELET

Telles sont, très résumées, les principales conclusions d’une importante étude du
Professeur Kees Ruijgh, telle qu’elle a été publiée dans les Actes du Colloque Pierre
Chantraine.

Je voudrais tenter de reprendre ici ses conclusions et de les appliquer à la composi­

tion formulaire de l’épopée grecque. Le nombre des emplois de κε(ν) et de ctv chez
Homère est très considérable. Malgré le caractère parfois fastidieux de quelques
dépouillements, l’étude de l’ensemble des attestations permet d ’aboutir à des con­
clusions d ’ordre divers, dont certaines seront fatalement hypothétiques.
On examinera les quatre points suivants:

1. la façon dont les aèdes et spécialement Homère en usaient avec les particules
κε(ν) et αν dans la composition formulaire.
2. l’emploi des particules au sein d ’expressions traditionnelles qui attestent des
indices dialectaux déterminés ou des formes attachées à un moment de l ’évolu­
tion du grec en général. Dans la même direction, on sera amené à s’interroger sur
l’appartenance à un dialecte de traits qui, jusqu’à présent, n’avaient aucune éti­
quette dialectale, alors qu’ils reviennent régulièrement dans le même contexte.
3. l'appartenance dialectale respective de κε(ν) et de âv dans la langue épique. Le
nombre considérable de leurs emplois permet une meilleure évaluation de l’im­
portance des différentes phases de composition de la tradition épique grecque,
surtout pour la période qui a précédé immédiatement celui que nous appelons
Homère.
4. la correspondance qui devrait a p r i o r i exister entre les emplois les plus anciens
des particules au point de vue sémantique et leur apparition dans des expressions
dont le grand âge serait garanti par la tradition.

Voilà donc les quatre questions que je voudrais évoquer ici. Parodiant le Général De
Gaulle sur un tout autre sujet, je serais tenté d ’ajouter 'V a s t e p r o g r a m m e ! ’. Bien
entendu, je n’aborderai qu’un nombre réduit de points qui me semblent révélateurs,
laissant l’ensemble du dossier à une éventuelle publication ultérieure. Je reprends
donc dans l’ordre les quatre questions que je viens d ’annoncer.

1° L’usage que les aèdes font de κε(ν) et de αν dans la composition formulaire

Première constatation qui est certainement valable pour d’autres particules ou en

général pour des mots très brefs tels que beaucoup de pronoms: les petits mots alter­
nent facilement au sein de formules ou d’éléments formulaires. C’est ainsi qu’on
verra des cas d’altemance de a l κε et de οφρα, qui présente la même coupe métrique
au sein de la même expression traditionnelle aux vers suivants:

Ψ 543 ώ Άχιλευ, μάλα τοι κεχολωσομαι, αι κ ε τ ε λ έ σ σ η ς

A 82 αλλά τε και μετόπισθεν έχει κότον, ο φ ρ α τ ε λ έ σ σ τ } .
 LES PARTICULES KE(N) ET A N DANS LES FORMULES 249

A 523 Ή ρη· έμ οί δ έ κ ε ταΰτα μ ελήσ εται, όφρα τελέσ σ ω '

Ailleurs et même plus souvent, la formule occupe un hémistiche de Phexamètre, alors

que des particules et des conjonctions sont accumulées dans le prem ier hémistiche.
C ’est ce qu’on trouve pour les formules:

- θάνατον κ α ί κηρας άλ ύξει, datis les vers:

β 352 = ε 387 διογενη ς Ό δυ σ εύς θ ά ν α τ ο ν κ α ι κ η ρ α ς ό λ ύ ξ α ς .

ρ 547 = τ 558 πά σ ι μάλ’, ούδέ κ έ τις θ ά ν α τ ο ν κ α ί κ η ρ α ς ό λ ύ ξ ε ι .
χ 66 ή φ εύγειν, ος κεν θάνατον κ α ί κ η ρ α ς ά λ ύ ξ η '

• φάνη ροδοδάκτυλος Ή ώς, dans les vers:

A 477 = Ω 788 ήμος 6' ή ρ ιγένεια φ ά ν η ρ ο δ ο δ ά κ τ υ λ ο ς Ή ώ ς ,

ψ 241 κ α ί νύ κ' όδυρομένοισι φ ά ν η ρ ο δ ο δ ά κ τ υ λ ο ς Ή ώ ς,
1 707 αύτάρ έ π ε ί κ ε φ α ν η κ α λ ή ρ ο δ ο δ ά κ τ υ λ ο ς Ή ώ ς ,

et ύπέκφυγον αίπύν όλεθρον, dans les vers:

1 286 αύτάρ έγώ συν τοίσ δε ύ π έ κ φ υ γ ο ν α ί π υ ν ό λ ε θ ρ ο ν

μ 446 έ σ ιδ έ ε ιν ού γάρ κ εν ύ π έ κ φ υ γ ο ν α ί π υ ν ό λ ε θ ρ ο ν .
μ 287 γίγνο ν τα ι- πη κ έν τις ύ π ε κ φ ύ γ ο ι α ί π υ ν ό λ ε θ ρ ο ν .

Toutes ces formules sont précédées d ’un hémistiche qui, dans plusieurs cas, contient
la particule κ ε. On note ainsi que κ ε(ν) et aussi â v tendent à être groupés avec
d ’autres particules ou avec des adverbes ou des conjonctions. L ’ensemble crée une
unité commode pour l ’aède qui place, de préférence ou même toujours, l’expression
au même endroit de l’hémistiche.
Ainsi, κ α ί νύ κε(ν), figure toujours en tête des vers suivants, à une exception
près (Θ 90):

E 311 κ α ί ν ύ κ ε ν ε ν θ ' ά π ό λ ο ι τ ο άναξ άνδρών Α ινείας,

Ε 388 κ α ί ν ύ κ ε ν ε ν θ ' ά π ό λ ο ι τ ο "Αρης άτος πολέμοιο,
Γ 373 = Σ 165 κ α ί ν ύ κ ε ν είρ υσ σ έν τε κ α ί άσπετον ήρατο κύδος,
λ 317 κ α ί vti κ ε ν έξετέλεσσαν, ε ί ήβης μέτρον ίκοντο'
Ψ 382 κ α ί ν ύ κ ε ν ή παρέλασσ' ή άμφήριστον έθηκεν,
Λ 311 κ α ί ν ύ κ ε ν έ ν νή εσσ ι πέσον φεύγοντες 'Αχαιοί,
Η 273 κ α ί ν ύ κ ε δη ξιφέεσσ' αύτοσχεδον ούτάζοντο,
Ρ 530 κ α ί νύ κ ε δη ξιφ έεσσ’ αύτοσχεδον όρμηθητην
φ 128 κ α ί ν ύ κ ε δή ρ’ έτά νυσ σ ε βίη το τέταρτον άνέλκω ν,
Θ 90 "Εκτορα· κ α ί ν ύ κ ε ν ενθ' ό γέρω ν από θυμόν όλεσ σ εν
 PAULWATHELET

Κ α ί, qui y a en général la valeur d’une conjonction appuyée, peut éventuellement

recouvrir un plus ancien κά ς, attesté en arcado-cypriote.8 Νυ est très probablement
un trait achéen de la langue épique.9 U est donc très possible que l ’ensemble soit
ancien dans la tradition et qu’il remonte à la phase de com position achéenne de
l ’épopée. L ’espace entre κ α ί νύ κε(ν) et la coupe médiane est occupé par (tes mots
divers. On y trouve encore des particules comme δη, ρα ou des adverbes comme έτι
ou ένθ α . Quelques termes sont susceptibles d ’être des éolismes com m e έτάνυσσε
(φ 128), έξετέλεσ σ αν (λ 317), εΐρ υσ σ εν (Γ 373 = Σ 165), avec un double -σ- con­
servé ou analogique,10 ou com m e dans ξ ιφ έεσ σ ι (H 273 et P 530) et νη εσ σ ι (Λ
311), avec le datif athématique éolien en -εσ σ ι.11
Autre exemple, l ’expression ε ίς ο κε(ν) occupe plusieurs emplacements dans le
vers, mais elle se trouve particulièrement après la diérèse bucolique et en tête du vers.
En plusieurs endroits, ε ί ς ö κ ε accompagne des formes de l ’aoriste second de
άφ ικνέομαι:

λ 122 ε ίς ο κ ε τους άφίκηαι, ο ϊ ού ίσ α σ ι θάλασσαν

et ψ 269 ε ίς ο κ ε τούς άφίκωμαι, ο ι ού ίσ α σ ι θάλασσαν
et aussi en Λ 193 κ τ είν ειν , ε ίς ο κ ε νηας έϋσσέλμους άφίκηται
Λ 208 κ τείν ειν, ε ίς ο κ ε νηας έϋσσέλμους άφ ίκηαι
Ρ 454 κ τείν ειν, ε ίς ο κ ε νηας έϋσσέλμους άφίκω νται
et Π 455 ε ίς δ κ ε δη Λ υκίης εύ ρείη ς δήμον ίκ ω νται

Après la diérèse bucolique, on trouve notam m ent plusieurs formes de l ’aoriste

ήλθον, dans les vers suivants:

o 543 = p 56 ένδυκέω ς φ ιλέειν κ α ί τιέμ εν, ε ίς δ κεν έλθω.

£ 77 ΰ ψ ι δ' έπ ' εύνάω ν όρμίσσομεν, ε ίς δ κ εν ελθη
Φ 231 Τρωσί πα ρεσ τάμ εναι κ α ί άμύνειν, ε ί ς δ κ εν έλθη
Κ 62 αύ θι μένω μετά τοισ ι, δεδεγμ ένος ε ί ς δ κ εν ελθης,

La majorité des dialectes grecs, dont l’arcado-cypriote, le thessalien et le béotien, ont

conservé la préposition έ ν originelle, construite avec l ’accusatif. L ’ionien-attique et

8 P. Wathelet, ‘Mycénien et grec d’Homère. 2. La particule κ α ί’. AC 33 (1964), p. 10-44. — C J.

Ruijgh, ‘Quelques remarques sur l'absence de καί et sur l’emploi des particules -qe et -de dans les
textes mycéniens'. Proc, of the Cambridge Coll, on Myc. Studies. Cambridge, 1966, p. 203; Études
sur la grammaire et le vocabulaire du grec mycénien. Amsterdam: Hakkert, 1967, p. 331-333, §
293; Autour de 'se épique'. Études de syntaxe grecque. Amsterdam: Hakkert, 1971, p. 181-181
9 C.J. Ruijgh, L'élément achéen dans la langue épique. Assen: Van Gorcum, 1957, p. 57-67. — p.
64: ‘Il est possible que νυ ait existé en éolien aussi, quoiqu’il n'y ait qu’une seule trace, outre
προταινί [en béotien) et le fragment lyrique mentionné ci-dessus [16.1, incerte utrius auctoris.
Lobet-Page], vu renforçant un impératif (Del.3, 523, 165) en béotien.’
^ P . Wathelet, Les traits éoliens dans la langue de l’épopée grecque, Rome: Ateneo, 1970, p. 304-

11 P. Wathelet, Les traits éoliens .... p. 252-265 et ‘Les datifs analogiques en -εσσι dans la tradi­
tion épique’. PEG 104 (1991), p. 1-14.
 LES PARTICULES KE(N) ET A N DANS LES FORMULES 251

le lesbien ont employé, pour l ’allatif, ε ις issu de ένς.12 Dans l ’épopée homérique,
ε ις constitue donc a p r i o r i un trait lesbien ou ionien, du moins dans les passages où
ε ί ς ne peut être rem placé par έν. C ’est ce qui se passe dans l ’expression qui nous
occupe, puisque ε ίς se trouve au temps fort. On pourrait certes évoquer, pour ει-, un
allongem ent m étrique ou restituer, à date plus ancienne, έ ν *yô κε(ν) ou èv hô
κ ε(ν),13 avec l ’aspiration qui fait position. Aucune des deux explications n’emporte
la conviction. S ’ils ne refusent pas les allongements métriques, les aèdes y ont plutôt
recours quand il n ’y a pas moyen de faire autrement. L a seconde explication n ’est
guère plus acceptable puisque, si on rem onte trop haut dans le temps, il faudrait
restituer non έ ν *yô κε(ν), mais bien έ ν *yo5 κε(ν),14 qui devient inscandable.
Comme κε(ν) n ’est pas employé en ionien, il vaut mieux voir dans ε ίς ö κ ε un élé­
ment formulaire lesbien.
Les deux exemples qui viennent d ’être cités semblent, ancien pour le premier et
relativement récent pour le second. Plusieurs autres expressions comme ώς κε, οφρα
κε(ν), οππως κε,15 όππότε κε, όππόιερός κε, ούδέ κε(ν), αιψ ά κε, e t c ., pourraient
être mentionnées ici. Tantôt elles comportent une indication dialectale ou qui est liée
au développement de la langue, tantôt elles n ’en attestent pas.

On retiendra de ce prem ier point que les aèdes ont tendance à faire alterner les par­
ticules et, d ’une façon générale, les mots très brefs au sein des expressions formu­
laires. Souvent aussi, ils les placent en dehors des formules en les groupant. D ’un
bref examen de deux groupes de particules ainsi constitués, il ressort que κ α ί νύ
κ ε(ν) est probablem ent ancien et même achéen, alors que ε ί ς ο κε(ν) relève sans
doute de l ’éolien d ’Asie.

2° Passons m aintenant au second point, l ’em ploi de κ ε (ν ) et de dv dans des

expressions traditionnelles et les liens que ces particules peuvent avoir avec des
formes qui doivent appartenir à une phase dialectale déterminée, sans que, jusqu’ici,
la chose ait été suggérée.
Déjà à la fin du point précédent, on a rencontré des emplois de κε(ν) anciens ou
liés à une forme lesbienne. La particule κε(ν) se retrouve régulièrem ent au sein de
plusieurs formules.
On en mentionnera quelques-unes:

12 P. Chantraine, Diet, ét., p. 326, s.v. εις et ές. — C.J. Ruijgh, ‘D’Homère aux origines proto-
mycéniennes de la tradition épique. Analyse dialectologique du langage homérique, avec un excur­
sus sur la création de l'alphabet grec’. Dans: Homeric Questions. Essays in Philology. Ancient His­
tory and Archaeology, including the Papers of a Conference organised by the Netherlands
Institute at Athens ( IS may 1993). Amsterdam: Gieben, 1995, p. 79-80. — A. Thumb-A. Scherer, II,
p. 108, § 257,5c pour le lesbien, et p. 283, § 313,3c pour l'ionien.
A. Thumb-A. Scherer, II, p. 330-331, § 337, 10. — C J. Ruijgh, Études .... p. 64-65 § 39, et
‘D’Homère aux origines. . p. 73-75.
14 II semble toutefois que les occlusives finales avaient disparu dès l'époque des tablettes mycéni­
ennes, puisqu’on trouve la graphie -wi-de (-Fi6e) et non **wi-de-te, que l ’on aurait attendue (C J.
Ruijgh, Éludes. . p. 43, § 19).
15 P. Wathelet, Les traits éoliens. . p. 294-295.
252 PAUL WATHELET

λ 358 = υ 316 κ α ί κ ε τό βουλοίμην, κ α ί κ ε ν π ο λ ύ κ έ ρ δ ι ο ν ε ϊ η ,

σ 166 π α ιδ ί δε κ εν ε ϊπ ο ιμ ι έπος, τ ό κ ε κ έ ρ δ ι ο ν ε ΐ η ,
Η 28 = υ 381 άλλ’ ε ί μοί τι πίθοιο, τ ό κ ε ν π ο λ ύ κ έ ρ δ ι ο ν ε ϊ η -
Ζ 410 πά ντες έφ ορμη θένιες- έμ οί δ έ κ ε κ έ ρ δ ι ο ν ε ϊ η
Ρ 417 πά σ ι χ ά νο ι- τό κ ε ν ήμιν άφαρ π ο λ ύ κ έ ρ δ ι ο ν ε ϊ η ,
Γ 41 κ α ί κ ε τό βουλοίμην, κ α ί κ ε ν π ο λ ύ κ έ ρ δ ι ο ν ή ε ν

Le com paratif κέρδιον ,16 au neutre, n ’a pas de positif connu, il repose sur le radical
du substantif τό κέρδος, l e g a i n , et il est uniquement d ’emploi épique et poétique. Il
semble s’inscrire dans la tendance connue en m ycénien et reprise dans la langue
épique et qui consiste à former des comparatifs ou des superlatifs sur des substan­
tifs: on connaît le cas de w a - n a - k a - t e - r o en m ycénien 17 et, chez Homère, ceux de
β α σ ιλ εύ τερ ο ς et β α σ ιλ εύτα τος. Κ έρ διο ν est nettement formulaire, il occupe la
m êm e place dans l ’hexamètre, toujours accompagné d ’une forme du verbe ε ίν α ι . 18
Dans les emplois qui viennent d ’être cités, κ έρ δ ιο ν est accom pagné de κ ε et de
l ’o ptatif pour m arquer le potentiel ou de l ’im parfait pour exprim er l ’irréel du
présent. Pour les emplois, plus fréquents, avec κ ε et l ’optatif, on pourrait se trouver
en présence d ’une formule ancienne .19
U ne expression parallèle apparaît avec le neutre de νεμεσσητός, q u i p r o v o q u e
l ' i n d i g n a t i o n (ou p r o m p t à s ’ i n d i g n e r ) , adjectif verbal du verbe ν ε μ ε σ ά ω j e
m ’i n d i g n e , avec un double -σ-, probablement analogique et qui a de grandes chances
d ’être un éolisme .20 Elle apparaît aux vers:

Γ 410 κ έΐσ ε δ' έγώ ν ούκ έίμ ι - ν ε μ ε σ σ η τ ό ν δ έ κ ε ν ε ϊ η -

Η 336 έ ξ εΰνης άνστάσα, ν ε μ ε σ σ η τ ό ν δ έ κ ε ν ε ϊ η .
Ω 463 όφθαλμοΰς ε ίσ ε ιμ ν ν ε μ ε σ σ η τ ό ν δ έ κ ε ν ε ϊ η
χ 489 εσταθ' έ ν ί μ εγά ρ ο ισ ι- ν ε μ ε σ σ η τ ό ν δ έ κ ε ν ε ϊ η .

Autre exemple de trait formulaire, après la diérèse bucolique, οττι κ ε ν ειπ ω (-ης,
-η), q u ’il faut lire, avant le début de la phase ionienne de la tradition, δττι κε

16 P. Chantraine, Diet, il., p. 519, r.v. κέρδος — Cf. ke-do-io génitif, peut-être de Κέρδος, nom
parlant d’un artisan à Pylos (PY Ua 158.1). C J. Ruijgh, éludes .... p. 120-121, § 99 et n, 105,
qui rapproche *κέρδω, κέρδος, κερδίων de έχθω, έχθος, έχθίων. On trouve deux emplois de la
même expression, mais avec âv:
E 201 » X 103, 1 228 άλλ' έγώ ο ύ πιθόμην — ή t" âv πολύ κέρδιον ήεν —
X 108 ώς έρέουσ ιν έμοί δέ τότ âv πολύ κέρδιον εϊη
17 E. Vilborg, ρ. 148, § 6 9 ,2 — Cf. C J. Ruijgh, Études .... p. 381-382, § 353 — P. Chantraine.
Gramm, horn., I, p. 259.
18 ήεν, qui marque ici l’irréel du présent, est probablement un trait ionien. Il s'agirait de l ’emploi
au singulier de Tancienne forme de la troisième personne du pluriel *ήσ·εντ Au singulier, on aurait
attendu ής < * est, attestée en divers dialectes, dont le lesbien. P. Chantraine, Gramm, kom., I, p.
288-289, et Morphologie historique du grec 2, p. 206-207.
19 Cf. plus loin 4°.
20 P. Chantraine, Diet, it., p. 743, s.v. νέμω. Sur le caractère éolien du double -a-, cf. P. Wathelet,
Les traits éoliens..., p. 304-307.
 LES PARTICULES KE(N) ET AiVDANS LES FORMULES 253

*(Ρείπω, avec ô m , de * y o d - k w id , qui, avec le double τ, a chance d ’être un éolisme ,21

sans q u ’on p u i s s e tout à fait exclure la possibilité d ’un trait achéen. L ’élément
formulaire apparaît aux vers:

B 361 οΰ to i απόβλητον έπος έσ σ εται, οτπ κ εν ε ΐ π ω .

Ε 421 Ζ εΰ πάτερ, ή ρά τί μοι κεχολώ σ εαι, ο τ η κ ε ν ε ΐ π ω ;
Ξ 190 ή ρ ά ν ΰ μοί τι πίθοιο, φίλον τ έ κ ο ς , ο τ η κ ε ν ε ΐ π ω ,
α 158 Ξ εΐνε φίλ', ή κ α ί μοι νεμ εσ ήσ εαι ο τ π κ ε ν ε ΐ π ω ;
α 389 Ά ντίνο', ε ί πέρ μοι κ α ί άγάσ σ εαι, ο τη κ εν ε ΐ π ω ,
β 25 = 161,229, ω 454 κέκ λυ τε δή νΰν μευ, 'Ιθακήσιοι, ο τ τ ι κ ε ν ε ΐ π ω '
τ 378 κ ή δ ε σ ιν άλλ' α γε νΰν ξ υ ν ίει έπος, ο τ η κ εν ε ΐ π ω '
τ4 0 6 γαμβρός έμ ό ς θυγάτηρ τε, τίθ εσ θ ' ονομ’, ο τ η κ ε ν ε ΐ π ω ·
υ 115 κρήνον νΰ ν κ α ί έμ οί δειλ ή έπος, ο τ η κεν ε ΐ π ω -
Α 294 ε ί δή σ οί πα ν εργον ύ πείξομ αι ο τ τ ι κ ε ν ε ΐ π η ς ·
Ω 92 ε ιμ ι μέν, ούδ' αλιον έπος έσ σ ετα ι, οττι κεν ε ϊ π ρ .

Parfois une formule dans laquelle figure la particule κε(ν) com porte un élément
dont on peut se dem ander s’il n’indique pas un apport dialectal déterminé.
C ’est notamment le cas pour toute une série de verbes au subjonctif. À l’origine,
le subjonctif a été caractérisé par l’usage d ’une voyelle alternante - d o - qui, dans les
form es thém atiques, se contractait avec la voyelle thém atique, pour donner une
voyelle longue. Les désinences semblent avoir été les désinences primitives origi­
nelles, celles qui sont devenues, en grec, ce que nous appelons les désinences
secondaires .22 Toutefois, de par la nature m êm e du mode, les temps du subjonctif
étaient ressentis com m e primaires. C ’est ce qui explique que certaines formes aient
été dotées de désinences “primaires”. Ainsi, en ionien-attique, à la deuxième et à la
troisième personnes du singulier des subjonctifs présent ou aoriste actifs, on a ajouté
un iota souscrit dans notre graphie: λύη apparaissait com m e le pendant de λ ύ ε ι et
λύης comme celui de λύεις.
D ’autres dialectes ont réagi dans le m êm e sens, mais avec des résultats dif­
férents. C ’est ainsi qu’on trouve dans la langue épique une troisièm e personne du
singulier en -σι (type έθέλησι), une deuxième personne en -θα (type έθέλησθα) et,
m êm e, mais plus rarem ent, une première personne en -μι (type έθ έλω μ ι ).23 À la
deuxième et à la troisième personnes, notre texte dote les formes d ’un iota souscrit,
qui n ’est pas original, mais inspiré des formes correspondantes de l ’ionien-atäque.
A la deuxième personne, la désinence -θα est la vieille désinence du parfait,
com m e l ’indiquent la comparaison et, en attique mêm e, l'archaïque ο ισ θ α , de
ο ιδ α .24 En ionien-attique, -θα ou -σθα a été surtout employé dans les temps secon-

21 P. Wathelet. Les traits éoliens..., p. 294-295.

22 Y. Duhoux, Le verbe grec ancien. Éléments de morphologie et de syntaxe historique, Lûuvain·
Ja-Neuve 1992; Peeters, p. 4 6 4 , § 384 et p. 475. § 390.
23 p. Chantraine, Gramm, horn., I, p. 461-463.
24 P. Chantraine, Gramm, horn., I, p. 469-470 — Y. Duhoux. Le verbe grec ancien, p. 474-475, §
389, et p. 481, §396.
254 PAUL WATHELET

daires: plus-que-parfait ήδησθα de οίδσ, et imparfait ήεισθα de ειμι j ’ir a i , έφησθα

de φημι j e d i s , ήσθα de είμί j e s u is . La désinence a ainsi pris une coloration secon­
daire. En revanche, en lesbien, elle a été employée au présent de l’indicatif de certains
verbes athématiques, comme le montrent des formes du type τίθησθα, φίλησθα,
φαισθα.25
Ainsi, l’éolien d’Asie ressentait plus le εχησθα comme une finale primaire. Π est
donc tentant, dans ces conditions, de se demander si les quelques formes de
deuxième personne du singulier en -σθα de subjonctifs présents et aoristes ne sont
pas, dans la langue épique, des traits éoliens.26 L’hypothèse est renforcée par des
emplois fréquents de tels subjonctifs avec la particule κε(ν).
La même question se pose pour la désinence -<π à la troisième personne et pour
la désinence -μι à la première. En ce qui concerne la troisième personne, la présence
de -σι, avec assibilation pour -τι, exclut une influence de l’éolien occidental.27 On
peut donc, a p r io r i, attribuer la finale -σι à l’achéen, au lesbien ou à l’ionien, mais le
cas parallèle de -σθα et l’usage fréquent de tels subjonctifs avec κε(ν) indiquerait
plutôt le lesbien. On trouvera, en annexe, la liste, aussi complète que possible, des
subjonctifs en -σθα, -σι et -μι, avec d’abord les formes en fin de vers, puis devant la
césure trochai'que, puis, enfin, dans d’autres positions.
On y voit clairement que l’extension des trois désinences procède du même
mouvement et que la langue épique emploie les trois, surtout avec les mêmes verbes.
C’est particulièrement le cas des formes verbales répétées comme celles de éfkAov,
de έθέλω, de ειπον, e t c . Même les formes isolées s’inscrivent dans le même sys­
tème de schémas formulaires.28
On se trouve donc en présence d ’un ensemble traditionnel dans lequel les sub­
jonctifs présents ou aoristes en -μι, -σθα et -σι constituent un système cohérent. La
particule κε(ν) y est très fréquemment insérée au point de faire corps avec lui.

Le second point de notre investigation a fait apparaître des formules d ’âges divers,
anciennes, peut-être achéennes ou beaucoup plus récentes, sans doute lesbiennes.
L’observation rejoint ce qu’on avait déjà noté pour les emplois de κε(ν) dans des
ensembles de particules. L’analyse formulaire a montré le lien entre l’emploi de
κε(ν) et les subjonctifs en -μι, -σθα et -σι. On a pu émettre l’hypothèse que de tels
subjonctifs constituaient des traits relevant de l’éolien d’Asie.

3° Le troisième point qu’il nous faut aborder est celui de l'appartenance dialectale
de κε(ν) et de ôv. En bref, κε est attesté en cypriote et en éolien, alors que άν appa-

25 A. Thumb-A. Scher«, Π, p. 100, § 256,13.

26 P. Wathelet.L« traits é o lie n s .... p. 311-312.
27 C f. J. Ruijgh, 'D’Homère aux origines ...', p. 5 — L'assibilation est un irait très ancien et
commun au proto-achéen et au proto-ionien-atlique. Si l’assibilation de -τι constitue un phénomène
assez banal dans beaucoup de langues, il est curieux que l'éolien l'ait connu beaucoup plus tard que
les autres dialectes grecs. Les changements phonologiques ne sont guère empruntés et on peut se
demander si le phénomène est en lesbien d’ordre phonologique ou d’ordre morphologique, en quoi
il serait lié seulement à l’emploi de certaines formes.
28 Cf. ci-après.
 le s p a rtic u le s ke (N) et an d a n s le s f o r m u le s 255

raît en arcadien et en ionien-attique .29 A p r i o r i , κε et âv pourraient donc appartenir à

la phase de composition achéenne de l ’épopée grecque, κ ε à la phase éolienne et âv
à la phase ionienne, c ’est-à-dire la dernière.
Comm e on l ’a remarqué depuis très longtemps, κ ε (ν ) est beaucoup plus
fréquent dans le texte homérique que âv . Pierre Chantraine signale quelque 1200
exemples de κε(ν) contre 350 environ de â v .30 Faut-il souligner que ces chiffies
sont approximatifs ? Un nombre exact ne peut être cité du fait que, pour certains vers,
la tradition manuscrite hésite entre κε(ν) et âv et surtout du fait qu’un nombre con­
sidérable de â v peuvent être remplacés p ar κ ε(ν) sans altérer le m ètre .31 D ’un
compte fatalement approximatif, il résulte qu’il ne reste q u ’une cinquantaine d ’em ­
plois de âv irréductibles en κε(ν), ce qui est évidemment très peu.
En général, les em plois irréductibles de âv sont peu formulaires, tout au plus
peut-on mentionner en fin de vers:

Γ 25 ε ι περ â v αύτόν
E 224 ε ι περ âv αυτε
Β 597 ε ι περ âv αύταί

Parmi les emplois de âv susceptibles d ’être remplacés par κε(ν), il faut certainement
se garder d'opérer partout la substitution et des éléments qui reviennent dans
plusieurs vers appartiennent sûrement au langage traditionnel. C ’est le cas de ώς 6 ’
οτ’ âv (ε 394, κ 216, 410, χ 468, ψ 233, Κ 5, Λ 269, Μ 4 1 ,0 80,170, Ρ 520, Τ 375,
Ω 480), en tête du vers, spécialem ent pour introduire une com paraison dite
homérique .32 On sait que ces comparaisons qui, par leur nature même, sont contem ­
poraines d ’Homère, contiennent une m ajorité d ’éléments récents. On rencontre
aussi, en plusieurs passages, l’expression οφρ’ âv, notamment dans οφρ' άν ϊκ η α ι (ζ
304, η 319), οφρ’ âv ϊκ ο ιο (κ 65), οφρ’ άν ϊκω μ αι (Κ 325, Φ 558), οφρ' άν ϊκ η τα ι
(Ο 23), chaque fois en fin d ’hexamètre. Toujours à la même place, le même verbe
apparaît avec κ ε , n o t a m m e n t e n Fin du vers A 139 ο ν κ ε ν ϊ κ ω μ α ι , de Z 225 οτε κεν
τω ν δήμον ϊκω μ αι, et de θ 32 οτις κ’ έμ ά δώμαθ' Ϊκητα ι, e tc .
Une autre constatation a été faite depuis longtemps sur l ’usage de âv dans notre
texte de l ’épopée: c ’est le nombre élevé des séquences ούκ άν que l ’on trouve en
diverses positions dans l ’hexamètre. J’en ai compté une soixantaine sur les quelque

29 Cf. plus haut, n. 4-7.

3®P. Chantraine, Gramm, ham ., II, p. 345.
31 Comme parexemple.cn μ 21311 νΟν δ' άγεθ', ώς άν έγώεΤπω, qui pourrait être lu vt)v 8 ' άγε
θ' ώς κεν éyà> εϊπω — Sur une constatation qui confirme que la substitution de âv à κε(ν) a dû
être opérée par Homère ou ses contemporains, cf. C J. Ruijgh, 'Laplace des enclitiques dans l’ordre
des mots chez Homère d'après la loi de Wackcmagcl'. Dans Scripta Minora II, Amsterdam: Gieben,
1996, p. 645-646 (= Sprachwissenschaft und Philologie. Jacob-Wackernagel und die Indo-
ermanistik heute, Wiesbaden: Reichen, 1990, p. 231-232)— Sur les quelques emplois conjoints de
t V et de κε, çf. C J. Ruijgh, ‘La place des enclitiques. . p. 233.
32 Les comparaisons homériques datent nécessairement de l'époque du poète lui-même et on a pu
montrer qu'en général leur langue est contemporaine du poète. Cf. G.P. Shipp, ‘Studies in the Lan­
guage of Homer’. Trans, o f the Cambridge Philol. Soc., 8 (Cambridge: 1953).
256 PAULWATHELET

350 attestations de άν. Π est évidemment tentant de remplacer ούκ αν par ού κεν, ce
que d ’aucuns ont fait. On a rappelé au début de la présente com m unication qu’av
avait pu apparaître p ar suite d ’une m écoupure de ού καν, lu ούκ αν. Le nombre
élevé de ούκ ôv dans l ’épopée homérique serait-il un indice qu’au moment où com­
m ence la phase de composition ionienne de l ’épopée, l’ionien en est encore à hésiter
entre κά ν et αν ? C ’est apparemment la situation qu’on retrouve beaucoup plus tard
(au Ve s. av. J.-C.) en arcaidien. L ’hypothèse mérite sans doute d ’être formulée, mais
la prudence recommande de ne rien fonder sur une base aussi mince. Π faudrait que
d ’autres constatations du même genre viennent la confirmer.

L ’épopée atteste encore un certain nombre de ήν, issus de ε ί αν, et de έπήν, issus de
έ π ε ί άν. Ces formes sont embarrassantes dans la mesure où elles com portent une
contraction .33 D existe une bonne vingtaine d ’emplois de ήν irréductibles en α ϊ κ’(ε),
en des endroits variés de l ’hexamètre, surtout en début de vers. En général, les con­
textes sont différents .34 En revanche, ήν peut être remplacé p ar α ϊ k' en six vers,
notamment dans l ’expression formulaire ήν έθέλησθα, susceptible de recouvrir un
plus ancien α ϊ κ' έθ έλη σ θα (Δ 353).
En ce qui concerne έπήν, on en trouve une quarantaine, dont la moitié dans l’ex­
pression αύτάρ έπήν, en début de vers .35 L a substitution de έ π ε ί κ' (ou χ') à έπήν
est possible chaque fois que έπήν est suivi d ’une voyelle, soit dans treize vers. La
substitution est encouragée par l ’existence de flottem ents dans la tradition
manuscrite entre έπήν et έπ ε ί χ ’, comme en χ 254.
Quoi qu’il en soit, les emplois irréductibles de ήν et de έπήν ne peuvent qu’ap­
partenir à la période la plus tardive de la composition épique.

Au début de ce paragraphe, j ’ai rappelé que άν apparaissait en arcadien et en ionien-

attique. L ’emploi de άν dans l ’épopée pouvait donc, a p r i o r i , être attribué à l ’achéen
ou à l’ionien. L ’examen, même rapide, de άν chez Homère m ontre que la particule
n ’apparaît dans aucune formule ancienne et que, pratiquement, l ’attribution de άν à
la phase de com position achéenne est des moins probables et m êm e exclue. En
d ’autres termes, un ά ν irréductible dans la langue épique devrait être considéré
comme un ionisme.
Non seulement, le nombre de άν attestés est beaucoup moins important que celui
de κε(ν)— la proportion est en gros de un pour quatre— , mais un grand nombre des
άν sont susceptibles de recouvrir un κε(ν) plus ancien. Même s ’il faut sans doute se
garder d ’opérer la substitution partout où elle est possible, il ne reste en définitive
qu’une cinquantaine de άν irréductibles, auxquels il faut ajouter quelques emplois
forcém ent récents de ήν et de έπήν. Comme Kees Ruijgh l ’a indiqué à plusieurs

33 P. Chantraine, Gramm, hom., I, p. 85, et II, p. 258-259 et p. 281-282 et p. 348.

34 Signalons seulement la finale ήν που άκούση (α 94) et ήν που άπούσω (β 360, γ 83), à côté de
plusieurs emplois de formes analogues de άκούω à la même place, mais dans d ’autres contextes.
35 Avec έπήν irréductible en a 293, γ 45, δ 412, ε 348, 363, ζ 262, 297, λ 119, ο 36, « 269, φ
159, χ 440.
 LES PARTICULES KE(N) ET A N DANS LES FORMULES 257

re prises ,36 si la langue d ’Homère est bien l ’ionien d ’Asie, la phase ionienne de
composition a commencé peu de temps avant l’époque du poète lui-même; l’apport
ionien est sans doute beaucoup plus limité que l ’apport éolien.

4° J’en viens maintenant à mon dernier point: y a-t-il une correspondance entre les
emplois les plus a n c i e n s de κ ε ( ν ) et s o n apparition dans des formules qui seraient
anciennes, elles aussi ?
Κε(ν) a été utilisé dans des emplois archaïques attestés par l’épopée homérique
e t qui n ’ont pas été conservés dans le grec ultérieur. Parm i les emplois les plus
archaïques, mentionnons celui du subjonctif avec κ ε dans la principale.
À l ’origine, le subjonctif avait une double valeur: ju ssif ou éventuel. Dans le
second cas, il se trouvait ainsi proche du futur en voie de formation. La valeur
éventuelle du subjonctif dans la principale a été soulignée par l’addition de la parti­
cule κε(ν), dont le sens primitif était a b r s , à c e m o m e n t f u t u r . 37
D ans l ’ensemble, les emplois de subjonctif avec κε(ν) dans la principale sont
peu formulaires. Ou bien, le verbe est isolé comme en

S 80 άνδρών δ' ή κ έν τις μοι έρ ίσ σ ετα ι, ή έ κα ί ούκί,

ou en Ξ 235 π είθ ευ , εγώ δ έ κ έ τοι είδ έω 38 χά ρ ιν ήματα πάντα,

ou, plus fréquem m ent, la particule κ ε n ’est pas liée à la formule à proxim ité de
laquelle elle apparaît, comme c ’est le cas en

Π 129 δύσεο τεύ χ εα θ ά σ σ ον έγώ δ έ κ ε λαόν άγείρω .

que l ’on rapproche de

Δ 377 ξειν ο ς άμ ’ ά ν τιθ έφ Π ο λ υνείκ εϊ, λαόν ά γ ε ίρ ω ν

Β 664 ά ίψ α δ έ νη ας έπηξε, πολυν δ' ο γ ε λαόν άγείρα ς
Λ 716 έννυ χο ς, ούδ' άέκοντα Π ύλον κότα λαόν ά γειρ εν,
γ 140 μύθον μυθείσθην, τοΰ ε ϊν ε κ α λαόν άγειραν.

En fait, seul est formulaire l’ensemble λαόν et une forme de άγείρω .

De même

δ 391 κ α ί δ έ κ έ το ι είπησ ι, διοτρεφές, α ί κ' έθέλησθα.

36 ‘D’Homère aux o r i g i n e s ρ. 91-92.

37 C J. Ruijgh, ‘L'emploi le plus ancien ...’, p. 78 et p. 83, et D’Homère aux o r i g i n e s p . 58-59.
38 C J . Ruijgh. ‘L ’emploi le plus ancien ...’, p. 78 — Non seulement ε ίδέω est une forme isolée
dans l’épopée homérique, mais elle pose de sérieux problèmes de scansion: il faut, ou bien faire la
contraction ou la synizèse, dans le -έω au temps fort, ou bien choisir la variante Ιδέα. Mais est-ce
encore une fam e de parfait?
258 Pa u l w a t h e l e t

dont le premier hémistiche alterne notamment avec celui de

Z 459 καί ποτέ τις εϊπησιν ίδών κατά δάκρυ χέουσαν

Ζ 479 καί ποτέ τις είπησι- πατρός γ' δδε πολλον άμείνων ν .Ι .
Η 87 καί ποτέ τις είπησι καί όψιγόνων ανθρώπων,
ζ 275 καί νύ τις ώδ' είπησι κακώτερος άντιβολήσας-
φ 324 μή ποτέ τις είπησι κακώτερος άλλος 'Αχαιών
δ 389 = κ 539 ος κέν τοι είπησιν οδόν καί μέτρα κελεύθου

Tous ces exemples attestent, sauf un (ζ 275), le respect du wau initial de είπησι.39
Le dernier montre un emploi de κεν, mais dans une relative à valeur éventuelle.
Un peu plus étroitement lié à la formule est κε employé avec καταλέξω toujours
en fin de vers. Καταλέξω peut être analysé comme indicatif futur ou comme sub­
jonctif aoriste.

1 262 ε ί δέ σύ μέν μευ άκουσον, έγώ δέ κ έ τοι καταλέξω

γ 80 εΐρεαι όππόθεν ε ίμ έν έγώ δέ κέ τοι καταλέξω.
ξ 99 εστ' αφενός τοσσοΰτον έγώ δέ κέ τοι καταλέξω.

Π existe toutefois d’autres emplois de καταλέξω, la grande majorité d ’entre eux se

trouvant en fin d’hexamètre, mais dans d’autres contextes.
Même si l’emploi de κε avec le subjonctif dans la principale est archaïque, les
quelques exemples qui viennent d’être donnés - et d’autres pourraient être mention­
nés - montrent qu’il ne s’agit pas du tout d’emplois figés dans des formules très
anciennes, mais qu’au contraire la particule κε entre dans un jeu de composition
encore vivant
Le subjonctif éventuel était proche du futur.40 Cette proximité était encore ren­
forcée par l’existence de formes homonymes susceptibles d’être des futurs de
l’indicatif ou des subjonctifs aoristes à voyelle brève (on vient de voir le cas de
καταλέξω). L’analyse formulaire des emplois de κε(ν) avec l’indicatif futur conduit
aux mêmes conclusions que pour le subjonctif avec κε(ν) dans une principale. Ou
bien les formes au futur ne se trouvent qu’une seule fois avec κεν comme c ’est le
cas de:

A 139 άξω ελώ ν ό δέ κεν κεχολώσεται ον κεν ϊκωμαι.

à côté de E 421 Ζεΰ πάτερ, ή ρά τί μοι κεχολώσεαι, οττι κεν εΐπω;
Y 301 μή πως καί Κρονίδης κεχολώσεται, α'ί κεν Ά χιλλεύς
ω 544 μή πώς τοι Κρονίδης κεχολώσεται εύρύοπα Ζευς.
etc.

39 P. Chantraine, Gramm, hom.. I, p. 133.

40 P. Chantraine, Morphologie historique du grec?, p. 245-248. — Y. Duhoux, Le verbe grec..., p.
437-456, § 359-378.
 LES PARTICULES KE(N) ET A N DANS LES FORMULES 259

Ou bien, les formes de futur avec κ ε qui n ’est pas toujours dans le même vers ne se
trouvent q u ’une seule fois dans l ’épopée homérique. C ’est le cas pour άπαλθησ-
εσ θ ον 41 en

0 404 = 418 ουδέ κεν έ ς δέκατους περιτελλομένους ενια υτούς

θ 405 = 419 ε λ κ ε ’ άπαλθήσεσθον, α κ εν μάρπτησι κερ α υνός-

Ou bien, κ ε est dans le même hémistiche que le verbe au futur, mais ce même verbe
est employé ailleurs sans κ ε. C ’est le cas en Δ 176 κ α ί κ έ τις ώδ’ έ ρ έ ε ι Τρώων
ύ περηνορεόντω ν, où la formule de prem ier hémistiche alterne avec ώς π οτέ τις
έρ έ ε ι (Δ 182, Z 462, H 91).
La m ême conclusion s’applique aux très rares em plois du futur de l’indicatif
avec αν (I 167, X 6 6 - dans ce dernier vers, έ ρ ύ ο υ σ ι doit être un futur - P.
Chantraine, G r a m m , h o m ., I, p. 452).
Les particules κε(ν) et dv ont aussi été employées avec l ’optatif pour exprim er
le potentiel. Cette construction n’appelle pas de com mentaire ici puisqu’elle a été
e m p l o y é e à travers toute l ’histoire d e la la n g u e grecque. En revanche, la langue
épique se distingue par le fait qu’on y trouve aussi κε avec l ’optatif dans l ’hypothé­
tique, usage que le grec ultérieur n ’a pas conservé. Ici encore les conclusions sont les
mêmes que pour le subjonctif et pour l ’indicatif futur:

- formes de verbes à l ’optatif avec κ ε dans le même hémistiche qui n ’apparaissent

qu’une seule fois, comme, par exemple, φάγοισι en

β 76 ε ΐ χ ' ΰ μ εΐς γ ε ψάγοιτε, τάχ' αν ποτέ κ α ι τίσ ις είη 42

- vers dans lesquels le verbe à l ’optatif est répété ailleurs, à la même place dans le
vers, mais sans κε(ν), comme en

X 351 οΰδ’ ε ί κ έ ν σ' αΰτόν χρυσ ω έρύσ α σ θαι άνώγοι

à côté de

θ 70 πάρ δ έ δέπας οινοιο, π ιέ ίν οτε θυμός άνώγοι.

π 141 π ίν ε κ α ί ήσθ', οτε θυμός έ ν ί στηθεσσιν άνώ γοι-
σ7 οϋ νεκ' άπ αγγέλλ εσ κε κιώ ν, οτε πού τις άνώ γοι-
Δ 263 εσ τηχ’ , ώς περ έμοί, π ιέ ε ιν , οτε θυμός άνώγοι.

41 Sur la forme, cf. P. Chantraine, Gramm, hom., I, p. 328, p. 446 et p. 477.

42 Autres exemples: A 60; H 387; I 141 = 283; I 445; N 288; X 220; Ψ 592; β 246; η 315; θ
353; V 389-390; p 223 sqq. et B 597 (avec dv) — Une autre répétition apparaît dans le premier
hémistiche de B 123 e \ περ γάρ κ' έθέλοιμεν, repris en θ 205.
260 PAUL WATHELET

Θ 189 οινόν τ' έγκεράσασα πιειν, οτε θυμός άνώγοι.43

Tout au plus peut-on mentionner une formule de premier hémistiche qui apparaît en
deux passages de l ’I l i a d e

E 273 ε ί τούτω κε λάβοιμεν, άροίμεθά κε κλέος έσθλόν,

θ 196 ε ί τούτω κε λάβοιμεν, έελποίμην κεν Αχαιούς

On y remarquera le duel τούτω, mais aussi le fait que κε reste bref devant λάβοιμεν
< *σλάβοιμεν.44 À vrai dire, la forme ne pourrait pas entrer autrement dans
l’hexamètre dactylique.
Un deuxième cas doit être évoqué ici, c’est celui de l’optatif employé pour
exprimer l’irréel du passé. Il s’agit d’un emploi archaïque, puisque, à l’époque
d’Homère et après lui, l’irréel est rendu par l’imparfait pour le présent et l’aoriste
pour le passé. On ne mentionnera ici qu’un seul exemple, qui a déjà été évoqué pour
d’autres raisons: celui de καί νύ κεν ένθ' άπόλοιτο aux vers

E 311 καί νύ κεν ενθ' άπόλοιτο άναξ άνδρών Αινείας,

Ε 388 καί νύ κεν ενθ' άπόλοιτο Ά ρης άτος πολέμοιο,

On a déjà signalé l’ancienneté très probable de l’expression καί νύ κεν. On se

trouve ici dans un système où le verbe et les précisions qui l’accompagnent se trou­
vent dans le premier hémistiche, tandis que le sujet, avec ses déterminants, constitue le
second hémistiche.
J’ai donc repris des κε(ν) employés dans des constructions considérées comme
archaïques et il faut avouer que l’enquête est plutôt décevante. Que ce soit pour les
attestations du subjonctif avec κε dans la principale ou du futur de l’indicatif avec la
même particule, on a vu que cette dernière entrait dans un jeu d ’altemances déjà évo­
qué au premier point de la présente recherche et qui n’inscrit pas le κε(ν) dans une
expression formulaire ancienne. Une constatation du même genre vaut pour les
emplois de κε avec l’optatif pour marquer le potentiel dans la subordonnée hypothé­
tique. Tout au plus a-t-on pu déceler une formule qui contient une expression sus­
ceptible d ’être ancienne et l’optatif avec κε utilisé pour exprimer l’irréel du passé.
Tout ceci implique que les constructions considérées comme archaïques, parce que
proches du sens premier de la particule et parce qu’elles n’ont pas survécu dans le
grec ultérieur, étaient encore vivantes dans la langue des aèdes au moins jusqu’à la
fin de la phase de composition éolienne de l’épopée et même jusqu’à l’époque
d ’Homère, les deux moments n’étant pas très éloignés l’un de l’autre.

43 Autres exemples: Z 49-50 = K 380-381 et Λ 135; T 321 et P 102; μ 345 et κ 420; τ 589-590 et
Ψ 894 — Faute de place, on ne pourra traiter en détail des problèmes que posent les relatives avec
κε et l ’optatif pour lesquels on pourrait arriver aux mêmes conclusions.
44 Sur l ’étymologie de λαμβάνω, cf. P. Chantraine, Diet, ét., p. 616, s.v. λαμβάνω — Le traite­
ment des groupes *σλ-, *σρ-, *σμ- et *σν- à l’initiale pose un problème dans la langue épique. Je me
propose dŸy revenir.
 LES PARTICULES KE(N) CTAJVDANS LES FORMULES 261

Les conclusions de cet examen, forcément un peu rapide, sont de divers ordres.
En ce qui concerne l ’appartenance dialectale des particules âv et κ ε dans la
langue homérique, on peut dire que κε(ν) appartient sûrement à la phase de compo­
sition éolienne et très probablement aussi à la phase achéenne plus ancienne, alors
que ctv constitue un ionisme.
Parmi les traits liés dans la tradition formulaire à l ’usage de κ ε, figurent les
subjonctifs présents ou aoristes seconds dotés des désinences -μ ι, -σθα et -σι. Ces
subjonctifs constituent un ensemble et divers indices suggèrent qu’il s’agit d ’un trait
propre à l’éolien d ’Asie.
L ’ensemble de la recherche a permis de mettre en évidence l ’importance de la
phase éolienne dans la constitution de l ’épopée grecque. Dans l’enthousiasme, bien
légitime, du développement du linéaire B, on a voulu parfois minimiser ou même faire
disparaître l’éolien de la langue épique .45 Si la base linguistique de l ’épopée
homérique est l ’achéen, et sans doute même un achéen antérieur à l ’époque des
tablettes ,46 l’apport éolien reste essentiel, spécialement celui de l ’éolien d ’Asie. J ’ai
déjà eu l'occasion de supposer ailleurs 47 que, si l ’épopée grecque s’était tout entière
réorganisée autour de la guerre de Troie, c ’était dû aux Éoliens d ’Asie, qui pouvaient
légitimement se considérer comme les héritiers des Achéens. Ceux-ci avaient assiégé
Troie com m e les Éoliens occupaient Lesbos et le rivage asiatique proche de la
Troade. D ’après tout ce que nous savons, l ’importance respective des phases éoli­
enne et ionienne de la composition épique indique que la phase ionienne avant
Homère a été très courte et que le poète a été un des premiers à introduire l ’ionien
dans la langue épique.
Je serais même tenté d ’aller plus loin: le caractère vivant de l’usage de κ ε, alors
que l ’ionien em ployait ctv, donne à penser que non seulem ent celui que nous
appelons Homère était un des plus anciens parmi les aèdes ioniens de l ’épopée, mais
qu’il avait encore une connaissance pratique de l’éolien d ’Asie.
Cette opinion repose sur des constatations faites sur plusieurs sujets. Non
seulement Homère connaît l ’usage de κε(ν) et s’en sert, sans que l ’emploi de la par­
ticule soit figé, mais il utilise avec brio des datifs éoliens en -εσσι, notamment dans
les comparaisons dites homériques48 et qui sont le reflet de la vie à son époque. Une
réflexion du m êm e genre pourrait s ’appliquer à l ’em ploi du duel qui a disparu en
ionien, mais qu’Homère emploie encore à bon escient, là où cela lui convient.49
Je tiens à préciser immédiatement, afin d ’éviter toute équivoque, que la connais­
sance pratique q u ’Homère avait de l’éolien d’Asie ne doit pas ramener à une opinion
professée anciennement 50 et selon laquelle le mélange d ’éolien et d ’ionien dans la

45 Telle était, en son temps, la thèse de K. Strunk, Die sogenannten Aeoiismen der homerischen
Sprache. Diss. Cologne. 1957.
46 C J. Ruijgh, ‘D'Homèreaux origines . .. \ p . 63-91.
47 P. Wathelet, Les Troyens de l’Iliade. Mythe et Histoire. Paris, 1989, p. 191.
48 P. Wathelet, Le temps des Héros et le temps du Poète. Univ. de Liège: Faculté ouverte, 1993.
49 Je compte revenir prochainement sur l’emploi du duel dans la langue épique.
50 La question est brièvement évoquée par P. Chantraine, Gramm, hom., I, p. 512.
langue épique viendrait du fait que l’épopée aurait été composée dans une région où
les deux dialectes étaient mêlés. On sait aujourd’hui que la réalité est plus com­
pliquée.
Une dernière remarque me semble s’imposer: nous suivons l’évolution de la
langue grecque à travers le développement de la tradition, en distinguant les faits
anciens et les faits récents. Or l’évolution même de la tradition épique brouille les
cartes. L’achéen était peut-être la première langue commune de la Grèce, langue fata­
lement évoluée, puis, par suite de circonstances qui n’ont rien de linguistique, vient
un apport éolien, c’est-à-dire l’apport d ’un dialecte qui était jusqu’alors marginal et
qui influence la langue de l’épopée, avant que la tradition de celle-ci revienne à
l’ionien, dialecte plus évolué. La conséquence pratique de cet état de chose est que
l’éolien a pu attester des faits plus archaïques que l’achéen, alors que la phase éoli­
enne est chronologiquement postérieure à la phase achéenne. Il résulte de cet état de
chose que la perspective est bouleversée et qu’il convient d’en tenir compte.

ANNEXE

O n trouvera ici, au tan t q u e p ossible, to u s les subjonctifs dotés des désin en ces -μ ι, -σθα , - σ ι. O n
c onstatera que, dans un nom bre c onsidérable d e c a s, su rto u t en fin d e v ers o u ap rès la césu re
trochaïque, les form es en ·μ ι, -σθα et -σι so n t accom pagnées d e κε. C ertaines form es, com m e S<5kn,
présentent des contractions qui sont récentes.

E n fin d e v e rs:

-μι

έθέλωμι φ 348 ιώ ν οΟ τις μ' άέκονχα βιήσεται αΧ κ' έθέλωμι

τύχωμι Ε279 νΰν αυτ' έγχείη πειρήσομαι, α ι κε τύχωμι.

-σθα

βάλησθα μ 221 κεισ' έξορμήσασα και έ ς κακόν άμμε βάλησθα.

βουλεύησθα I 99 σκηπτρόν τ' ήδέ θέμισχας, Χνα σφίσι βουλεύησθα.
έθέλησθα Σ 457 = γ 92, 5 322 χοΰνεκα ν€ν τά σά γούναθ' ίκάνομαι, αί κ' έθέλησθα
5 391 καί δέ κέ τοι εΧπησι, διοτρεφές, αΧ κ' έθέλησθα
μ 49 χών άλλων άτάρ αύχός άκούεμεν, αΧ κ' έθέλησθα,
υ 233 σοισιν 5' όφθαλμοΧσιν έπόψεαι. α ι κ' έθέλησθα.
Κ 235 τόν μέν δη έχαρόν γ ’ αίρήσεαι, Öv κ' έθέλησθα.
HÉS. Op. 392 γυμνόν 5' όμάειν, εΧ χ ’ ώρια πανί' έθέλησθα
Φ 484 Ζευς θήκεν, και έδωκε κατακτάμεν, ην κ’ έθέλησθα
 LES PARTICULES KE(N) ETA N DANS LES FORMULES 263

Ω 335 άνδρί έταιρίσσαι, καί τ' έκλυες φ κ' έθέλησθα,

A 554 άλλα μάλ' εύκηλος τά φράζεαι άσσ' έθέλησθα.
έχησθα Τ 180 πιείρη, ϊνα μή τι δίκης έπιδευές έχησθα.
πάθησθα Ω 551 ούδέ μιν άνστήσεις, πριν καί κακόν άλλο πάθησθα.
πίησθα Ζ 260 πρώτον, έπειτα δέ καυτός όνήσεαι, αί κε πίησθα.

-σι

δγησι ζ 37 ήμιόνους καί άμαξαν έφοπλίσαι, ή κεν άγησι

άποπέσησιν ω 7 τρίζουσαι ποτέονται, έπεί κέ τις άποπέσησιν
βάλησι(ν) Φ 103 δν κε θεός γε
Φ 104 Ιλίου προπάροιθεν έμής έν χερσί βάλησι,
Φ 576 εί περ γάρ φθάμενός μιν η ούτάση ήέ βάλησιν,
έθέλησι θ 142 σήμερον ύστερον άντε καί ήμΐν, αί κ' έθέλησι.
Κ 55 είμι, καί ότρυνέω άνστήμεναι, αί κ' έθέλησιν
Σ 143 είμι παρ' "Ηφαιστον κλυτοτέχνην, αί κ' έθέλησιν
Σ 278 στησόμεθ’άμ πύργους- τφ δ’άλγιον, αί κ' έθέλησιν
β 128 πριν γ' αύτήν γήμασθαι 'Αχαιών φ κ' έθέλησι.
κ 22 ήμέν παυέμεναι ήδ' όρνύμεν δν κ' έθέλησι.
ρ 11 δαίτα πτωχεύη· δώσει δέ οί δς κ' έθέλησι
ρ 19 δαίτα πτωχεύειν· δώσει δέ μοι δς κ' έθέλησιν.
ρ 559 γαστέρα βοσκήσεις· δώσει δέ τοι δς κ' έθέλησιν.
φ 280 ήώθεν δέ θεός δώσει κράτος φ κ' έθέλησιν.
HÉS. Théog. 430 έν τ’ άγορή λαοίσι μεταπρέπει δν κ’έθέλησι
432 άνέρες, ένθα θεά παραγίγνεται οΧς κ' έθέλησι
439 έσθλή δ' ίππήεσσι παρεστάμεν οίς κ' έθέλησιν
έλησι μ 96 δελφίνος τε κύνας τε καί εί ποθι μείζον έλησι
ρ 323 άνέρος, εύτ' άν μιν κατά δούλιον ήμαρ έλησιν.
έρέθησιν A 519 "Ηρη. δτ’ άν μ’ έρέθησιν όνειδείοις έπέεσσι'
έρίπησι Ρ 522 ΐνα τάμη διά πάσαν, ό δέ προθορών έρίπησιν
δχησι(ν) α 95 = γ 78 ήδ’ΐνα μιν κλέος έσθλόν έν άνθρώποισιν έχησιν.
α 204 έσσεται, οϋδ' εί πέρ τε σιδήρεα δέσματ' έχησι·
fai Ο 359 γίγνεται. όππότ' άνήρ σθένεος πειρώμενος fai.
δ 756 έχθέσθ', άλλ' έτι πού τις έπέσσεται δς κεν έχησι
θάνησι(ν) Τ 228 άλλά χρή τόν μέν καταθάπτειν δς κε θάνησι,
λ 218 άλλ' αδτη δίκη έστί βροτών, δτε τίς κε θάνησιν
θέησιν Σ 601 έζόμενος κεραμεύς πειρήσεται, αί κε θέησιν
ΐησιν I 701 άλλ' ήτοι κείνον μέν έάσομεν, ή κεν ΐησιν,
κίχησι μ 122 δείδω μή σ' έξαΰτις έφορμηθεισα κίχησι
λάβησιν Φ 24 δειδιότες· μάλα γάρ τε κατεσθίει δν κε λάβησιν
Ω 43 είξας είσ' έπί μήλα βροτών. ΐνα δαίτα λάβησιν
α 192 παρτιθεί, εύτ' άν μιν κάματος κατά γυία λάβησιν
λάθησι μ 220 νήα. σύ δέ σκοπέλου έπιμαίεο, μή σε λάθησι
 PAULWATHELET

λάχησιν Η 17] κλήρφ vOv πεπάλασθε διαμπερές, δς κε λάχησιν

μένησι X 93 ώς δέ δράκων έπι χειή όρέστερος άνδρα μένησι,
πάθησι(ν) Ρ242 δσσον έμή κεφαλή περιδείδια, μή τι πάθησι,
Τ 126 τησδε μάχης, ΐνα μή τι μετά Τρώεσσι πάθησι
δ 820 τοΰ δ' άμφιτρομέω και δείδια μή τι πάθησιν,
η 195 μηδέ τι μεσσηγύς γε κακόν και πήμα πάθησι
πέσησι(ν) 0 624 έν δ' έπεσ' ώς δτε κύμα θοή έν νηΐ πέσησι
θ 524 δς κε έης πρόσθεν πόλιος λαών τε πέσησιν,
λ 254 τών δ' άλλων οϋ κήδος, έπεί χ" οΰτός γε πέσησιν.
πόρησιν β 186 σφ οΐκφ δώρον ποτιδέγμενος, αΐ κε κόρησιν.
π 77 μνάται évi μεγάροισιν άνήρ και πλείστα πόρησιν.
υ 335 γήμασθ' δς τις δριστος άνήρ και πλείστα πόρησιν,
προφέρησι I 323 ώς δ' δρνις άπτησι νεοσσοΐσι προφέρησι.
τάμησιν σ 339 κεΐσ' έλθόν, ΐνα σ' αΰθι διά μελεΐστ'ι τάμησιν.
φάγησιν θ 477 κηρυξ, τή δή, τούτο πόρε κρέας, δφρα φάγησι,
φέρησιν κ 507 ήσθαι' την δέ κέ τοι πνοιή Βορέαο φέρησιν.
φιλέησιν ο 70 άνδρΐ ξενοδόκφ, δς κ' έξοχα μέν φιλέησιν,
φύγησι Ε 258 άμφω άφ' ήμείων, εΐ γ' ούν έτερός γε φύγησιν.
Η 118 φημί μιν άσπασίως γόνυ κάμψειν, αΐ κε φύγησι
Η 173 και δ’ αύτός δν θυμόν όνήσεται, αΐ κε φύγησι
Τ 72 άσπασίως αύτών γόνυ κάμψειν, δς κε φύγησι

Devant la césure ι

-μι

έθέλωμι I 397 τάων ήν κ' έθέλωμι φίλην ποιήσομ' άκοιτιν.

εΐπωμι X392 δφρα έπος εΐπωμι τό μοι καταθύμιόν έστιν
ΐδωμι Σ63 άλλ’ εΐμ’, δφρ' ΐδωμι φίλον τέκος, ήδ' έπακούσω
ΐκωμι I 414 εί δέ κεν οΐκαδ' ΐκωμι φίλην ές πατρίδα γαΐαν,
τύχωμι χ 7 εΐσομαι,
εΐ αΐ κε τύχωμι, πόρη δέ μοι εύχος 'Απόλλων

-σθα

δηθύνησθα μ 121 ήν γάρ δηθύνησθα κορυσσόμενος παρά πέτρη

έθέλησθα θ 471 δψεαι, αΐ κ' έθέλησθα, βοώπις πότνια Ήρη
Ν 260 δούρατα δ’, αΐ κ' έθέλησθα, και έν και είκοσι δήεις
Τ 147 δώρα μέν αΐ κ' έθέλησθα παρασχέμεν, ώς έπιεικές
ω 511 δψεαι, αΐ κ' έθέλησθα, πάτερ φίλε, τφδ' έπι θυμφ
I 288 τάων ήν κ' έθέλησθα φίλην άνάεδνον δγεσθαι
σ 270 γημασθ1φ κ' έθέλησθα, τεόν κατά δώμα λιποΰσα.
I 359 δψεαι, αΐ κ' έθέλησθα και αΐ κέν τοι τά μεμήλη,
 LES PARTICI)LES KE(N) ET A N DANS LES FORMULES 265

Δ 353 1δψεοι, ήν έθέλησθα καί αί κέν m τα μεμήλη,

είπησθα Τ 250 όπ
( ποιόν κ' εΐπησθα έπος, τοίόν κ’έπακούσαις.

-σι

άγάγησιν Ω 155= 1
άγησι σ 137 otovc έπ' ήμαρ άγησι πατήρ άνδρών χε θεών τε.
άγνοιήσι ω 218 τήέ κεν άγνοιήσι πολΰν χρόνον άμφίς έόντα.
άνέχησι τ 111 εύδικίας
i άνέχησι, φέρησι δέ γΟϊα μέλαινα
άποστρέψησιν Ο 62 αύτις άποστρέψησιν άνάλκιδα φύζαν ένόρσας,
δώησιν A 324 εί ί δέ κε μή δώησιν, έγώ δέ κεν αυτός έλωμαι
Μ 275 (αί κε Ζευς δώησιν ’Ολύμπιος άστεροπητής
έθέλησι(ν) Σ30ί6 άλγιον, αί κ' έθέλησι, τφ έσσεταΓ οϋ μιν έγωγε
A 408 cαί κέν πως έθέλησιν έπί Τρώεσσιν άρήξαι,
A 580 tεί περ γάρ κ' έθέλησιν Ολύμπιος άστεροπητής
δφσι α 379 = β■144 αί κέ ποθι Ζευς δφσι παλίντιτα έργα γενέσθαι-
έησι Β 366 τήδ' δς κ' έσθλός έηστ κατά σφέας γάρ μαχέονται-
έθέλησι(ν) 1 429 αύριον,
( ήν έθέλησιν ανάγκη δ' οΰ τί μιν άξω
I 692 αύριον, ήν έθέλησιν άνάγκη 5' ού τί μιν άξει.
0 21 0 νεικείειν έθέλησι- χολωτοΐσιν έπέεσσιν.
Ψ 554 άνδρών δς κ' έθέλησιν έμοί χείρεσσι μάχεσθαι.
σ 286 δώρα μέν δς κ' έθέλησιν 'Αχαιών ένθάδ' ένείκαι,
I 146 τάων ήν κ’ έθέλησι φίλην άνάεδνον άγέσθω
σ47 τάων ήν κ' έθέλησιν άναστάς αύτός έλέσθω-
Τ 243 δππως κεν έθέλησιν ό γάρ κάρτιστος άπάντων.
έθέλησ' αύτός δ', αί κ' έθέλησ', ίήσεται, ούδέ τις άλλος
HÉS. Ορ. 268 καί νυ τόδ’ . αί κ' έθέλησ', έπιδέρκεται, ούδέ έ λήθει
είπησι(ν) δ 389 = κ 539 δς κέν τοι είπησιν όδόν καί μέτρα κελεύθου
δ 391 καί δέ κέ τοι είπησι, διοτρεφές, αί κ' έθέλησθα
α 282 ήν τίς τοι είπησι βροχών, ή δσσαν άκούσης
β 216 ήν τίς μοι είπησι βροχών, ή δσσαν άκούσω
Ζ459 καί ποτέ τις είπησιν ίδών κατά δάκρυ χέουσαν
Ζ479 καί ποτέ τις είπησι' πατρός δ' δδε πολλόν άμείνων ν.Ι.
Η 87 καί ποτέ τις είπησι καί όψιγόνων ανθρώπων,
X 106 μή ποτέ τις είπησι κακώτερος άλλος έμείο-
Ψ 575 μή ποτέ τις είπησιν 'Αχαιών χαλκοχιτώνων
0 57 έλθη, καί είπησι Ποσειδάωνι άνακτι
Η 300 δφρα τις ώδ' είπησιν 'Αχαιών τε Τρώων τε'
ζ 275 καί νύ τις ώδ' είπησι κακώτερος άντιβολήσας'
φ 324 μή ποτέ τις είπησι κακώτερος άλλος 'Αχαιών
εισήλθησι θ 522 μή λόχος είσέλθησι πόλιν λαών άπεόντων.
έκπέμψησι σ 336 δώματος έκπέμψησι φορύξας άίματι πολλφ.
έκτάμνησιν Γ 62 νήϊον έκτάμνησιν, όφέλλει δ' άνδρός έρωήν
266 PAUL WATHELET

έλησι Ψ 345 ούκ έσθ' δς κέ σ' έλησι μετάλμενος ούδέ παρέλθη,

έλθησι(ν) Τ 191 έκ κλισίης έλθησι και δρκια πιστά τάμωμεν.
α 77 νόστον, δπως έλθησι· Ποσειδάων δέ μεθήσει
λ 192 αύτάρ έπήν έλθησι θέρος τεθαλυΐά τ’ όπώρη,
ξ 398 ει δέ κε μή έλθησιν άναξ τεός ώς αγορεύω,
ξ 515 = ο 337 αύτάρ έπήν έλθησιν Όδυσσήος φίλος υιός,
έμπνεύσησι Ο 60 αΰτις 5' έμπνεύσησι μένος, λελάθη δ' όδυνάων
έπέλθησιν Ω 651 ένθάδ' έπέλθησιν βουληφόρος, οί τέ μοι αίεί
έξερέησι δ 337 = ρ 128 κνημούς έξερέησι και άγκεα ποιήεντα
έπισσείησιν Δ 167 αυτός έπισσείτισιν έρεμνήν αιγίδα πάσι
εΰδησι Ε 524 άτρέμας, δφρ' εΰδησι μένος Βορέαο καί άλλων
εϋρησι Μ 302 εί περ γάρ χ' εύρησι παρ' αύτόφι βώτορας άνδρας
έχησι(ν) α 396 τών κέν τις τόδ' έχησιν, έπεί θάνε δΐος Όδυσσεύς·
υ 85 νύκτας δ' ύπνος έχησιν - ό γάρ τ' έπέλησεν απάντων,
έχθαιρησι(ν) δ 692 άλλον κ' έχθαίρησι βροτών, άλλον κε φιλοίη.
ο 71 έξοχα 5' έχθαίρησιν άμείνω δ' αΐσιμα πάντα,
θάνησι δ 196 κλαίειν δς κε θάνησι βροτών καί πότμον έπίσπη.
θέησι X 23 δς £>άτε βεία θέησι τιταινόμενος πεδίοιο·
ίλήκησι φ 365 ήμίν ίλήκησι καί άθάνατοι θεοί άλλοι·
καθέλησι β 100 = γ 238, τ 145, ω 135 μοιρ' όλοη καθέλησι τανηλεγέος θανάτοιο,
κάμησι Ρ 658 δς τ' έπεί άρ κε κάμησι κύνας τ' άνδρας τ' έρεθίζων,
ξ 65 δς οί πολλά κάμησι, θεός δ' έπί έργον άέξη,
κλαίησι θ 523 ώς δέ γυνή κλαίησι φίλον πόσιν άμφιπεσοΰσα,
λάβησι I 324 μάστακ', έπεί κε λάβησι, κακώς δ' άρα οί πέλει αυτή
λάθησι X 191 τόν δ', εί πέρ τε λάθησι καταπτήξας ΰπδ θάμνω,
μετέησιν Γ 109 οίς δ' 6 γέρων μετέησιυ, άμα πρόσσω καί όπίσσω
ότρύνησι(ν) 0 59 "Εκτορα δ’ότρύνησι μάχην ές Φοίβος 'Απόλλων,
ξ 374 έλθέμεν ότρύνησιν, δτ' άγγελίη ποθέν έλθη.
πάθησι(ν) X 505 νύν δ’ άν πολλά πάθησι, φίλου άπδ πατρός άμαρτών,
Λ 470 δείδω μή τι πάθησιν ένί Τρώεσσι μονωθείς,
παμφαίνησι Ε 6 λαμπρόν παμφαίνησι λελουμένος Ώκεανοίο'
παρεκπροφύγησιν Ψ 314 παντοίην, ινα μή σε παρεκπροφύγησιν άεθλα.
παυσήσι Δ 191 φάρμαχ', ά κεν παύσησι μελαινάων όδυνάων. [κεν /περ} ν.Ι.
βαίησι ε 221 εί δ' αύ τις βαίησι θεών ένί οίνοπι πόντω,
στείχησι λ 17 ούθ' όπότ' άν στείχησι πρός ούρανόν άστερόεντα,
στυγέησιν V 400 έσσω δ κε στυγέησιν ίδών άνθρωπος έχοντα,
τάμησι σ 86 δς κ' άπό βίνα τάμησι καί ούατα νηλέ'ί χαλκώ,
τύχησι Λ 116 ή δ' εί πέρ τε τύχησι μάλα σχεδόν, ού δύναταί σφι
φάγησι Φ 127 ιχθύς, δς κε φάγησι Λυκάονος άργέτα δημόν.
φέρησι(ν) Ζ 308 στήσομαι, ή κε φέρησι μέγα κράτος, ή κε φεροίμην.
ε 164 ύψοΰ, ώς σε φέρησιν έπ' ήεροειδέα πόντον,
φθήσιν Ψ 805 όππότερός κε φθήσιν όρεξάμενος χρόα καλόν,
φορέησι ι 10 οίνοχόος φορέησι καί έγχείη δεπάεσσι.
 LES PARTICULES KE(N) FT A N DANS LES FORMULES 267

φρονέησι η 74 οίσίν τ’ εύ φρονέησι και άνδράσι νείκεα λύει.

φύγησι Φ 296 Τρωικόν, δς κε φύγησι· σύ δ' Έκτορι θυμόν άπούρας
χραίσμησι Λ 387 ούκ άν τοι χραίσμησι βιδς και ταρφέες ιοί'

A utres positions:

•μι

άγάγωμι Ω 717 άσεσθε κλαυθμοΐο, έπήν άγάγωμι δόμονδε.

έθέλωμι A 549 δν 5έ κ' έγών άπάνευθε θεών έθέλωμι νοήσαι.
κτείνωμι τ 490 δμφάς έν μεγάροισιν έμοίς κτείνωμι γυναίκας.

είπησθα λ 224 ίσθ'. ινα και μετόπισθε τεή είπησθα γυναικί.

X373 δφρα γνφς κατά θυμόν, άτάρ είπησθα και άλλψ,
εύδησθα θ 445 εύδησθα γλυκΰν ύπνον ιών έν νηΐ μελαίνη.
παρεξελάσησθα Ψ 344 εί γάρ κ’ έν νύσση γε παρεξελάσησθα διώκων,
σπένδησθα 6 591 δώσω καλόν άλεισον, ινα σπένδησθα θεοΐσιν

-σι

άείδησιν τ 519 καλόν άείδησιν έαρος νέον Ισταμένοιο,

άλάλκησι κ 288 έρχευ, δ κέν τοι κρατός άλάλκησι κακόν ήμαρ.
άπέησιν τ 169 Ας άπέησιν άνήρ τόσσον χρόνον δσσον έγώ νύν,
άποδφσιν θ 318 εις δ κέ μοι μάλα πάντα πατήρ άποδφσιν έεδνα,
βρίθησι τ 112 πυρούς καί κριθάς, βρίθησι δέ δένδρεα καρπφ,
διαιρίβησιν β 204 έσσεται, δφρα κεν ή γε διατρίβησιν 'Αχαιούς
δώσι A 129 δφσι πόλιν Τροίην εύτείχεον έξαλαπάξαι.
έγειρησιν Κ 511 μή πού τις καί Τρώας έγειρησιν θεός άλλος.
Ζ 281 αί κ' έθέλησ' είπόντος άκούεμεν ώς κέ οί αΰθι
έθέλησ'(ι)(ν)
Ν 743 αί κ' έθέλησι θεός δόμεναι κράτος, ή κεν έπειτα
Ο 492 ήδ’ δτινας μινύθη τε καί ούκ έθέλησιν άμύνειν,
Π 53 όππότε δή τον όμοίον άνήρ έθέλησιν άμέρσαι
Π 446 φράζεο μή τις έπειτα θεών έθέλησι καί άλλος
α 349 άνδράσιν άλφηστήσιν, δπως έθέλησιν, έκάστφ.
ζ 189 έσθλοίς ήδέ κακοΐσιν, δπως έθέλησιν, έκάστφ·
ειπησι φ 229 έξελθών μεγάροιο, άτάρ ειπησι καί εϊσω.
έλησι τ 515 αύτάρ έπεί νυξ έλθη, έλησί τε κοίτος άπαντας,
έλθησιν γ 422 έλθησιν, έλάση δέ βοών έπιβουκόλος άνήρ·
έλθησ' Ε 132 = 821 έλθησ' ές πόλεμον, τήν γ' ούτάμεν όξέί χαλκφ.
έλκησιν Ψ 518 έλκησιν πεδίοιο τιταινόμενος σύν δχεσφι-
έπαγγείλησι δ 775 πάντας όμώς, μή πού τις έπαγγείλησι καί εϊσω
 PAUL WATHELET

έπιπνείησιν δ 357 ήνυσεν, ή λιγύς ουρος έπιπνείησιν όπισθεν

έποτρύνησι(ν) Π 690 ρηϊδίως, οτε δ' αύτός έποτρύνησι μάχεσθαι
θ 45 τέρπειν, 0ππη θυμός έποτρύνησιν άείδειν.
κλίνησι Τ 223 άμητος δ' όλίγιστος, έπην κλίνησι τάλαντα
μάρπτησι θ 405 = 419 ελκε' άπαλθήσεσθον, à κεν μάρπτησι κεραυνός'
μενοινήησι Ο 82 ένθ' εΐην, ή ένθα, μενοινήησι te πολλά,
νεικείησ ι A 579 νεικείησι πατήρ, σύν δ' ήμίν δαΐτα ταράξη·
παραιπεπίθησιν χ 213 Μέντορ, μή σε έπεσσι παραιπεπίθησιν Όδυσσεύς
πέμπησιν 0 109 τώ έ χ ε θ ', δττι κεν ϋμμι κακόν πέμπησιν έκάστφ.
πρησσησιν γ 476 ζεύξαθ' ύφ' άρματ άγοντες, ινα πρήσσησιν όδοΐο,
προθέησι Κ 362 χώρον άν' ύλήενθ', ό δέ τε προθέησι μεμηκώς,
στείχησ' η 72 δειδέχαται μύθοισιν, δτε στείχησ’ άνά άστυ.
στυγέησι Θ 515 νηός έπιθρφσκων, ινα τις στυγέησι καί άλλος
ύφαίνησιν ε 356 "Ω μοι έγώ, μή τίς μοι ύφαίνησιν δόλον αΰτε
φέρησι τ 111 εύδικίας άνέχησι, φέρησι δε γαια μέλαινα
φήσιν φήσιν έλεύσεσθαι' τοΟ δ' ώλετο νόστιμον ήμαρ.
φορέησιν ώς δ' όι' όπωρινός Βορέης φορέησιν άκάνθας
φρονέησι 313 ε ί κέν τοι κείνη γ ε φίλαφρονέησ' έ νί θυμφ.
 PARTICULARS OF SOME PROTO-INDO-EUROPEAN
PARTICLES

F.M.J. Waanders
Universiteit van Amsterdam

0. Introduction

After some initial reluctance towards atomizing small words such as prepositions and
other relators 1 have gradually becom e fascinated by the idea, and I am now con­
vinced that we are indeed dealing with conglomerates of primitive particles in the case
o f several historically attested small, non-nominal and non-verbal words— and even
endings— containing more than one Proto-Indo-European consonant. M y impression
is that any given consonant + vowel * e /o , when in its full-grade form, was in fact
used as a particle at an archaic stage of Proto-Indo-European .1 Those primitive parti­
cles, sometimes combinations o f two or more of them, appear to have provided the
basis for future inflexional suffixes, adpositions, adverbs, conjunctions, particles in
the customary sense, and pronominal forms, perhaps also one or two nominal or ver­
bal stems .2 It goes without saying that hypotheses concerning archaic PIE are at best
tentative, and often speculative— but I simply cannot resist the temptation.

1. * b ke ! o ‘(A) with (B)’

C onsider * b he / o . Originally, this particle perhaps expressed close association

between tw o entities, styled ‘réunion de l’un et de l ’autre’ by B ader (1981: 41),
though in several contexts this basic ‘1 + 1 ’-value was lost. * b he ! o lies at the basis
o f words for ‘both’, e.g. Gothic b a i , Sanskrit u - b h à u , Greek άμ-φω, Latin a m - b ô ,
and also o f the Germanic prepositions Gothic b i , Germ an b e i , Dutch b i j , and the
flexional endings Greek -φι, Sanskrit - b h ih and - b h y a h , Latin - b u s (and - b i in t ib l,
sibT ); it is an ingredient of Greek άμ-φί, Latin a m - b - in a m b i r e , Germ anic * u m b i
(OHG u m b i ) , Sanskrit a - b h i. Note that for one school o f Indo-Europeanists, which I
belong to, the - i o f * b hi is the syllabic allophone o f * y , which means that * b h i is
com posed o f the zero grades of * b he and * y e , o r even * b he , * h ,e and * y e . I am not

1 Except *r, which apparently did not occur in word-initial position in PIE. The archaic stage of
PIE referred to here is pre-inflexional ‘Early Indo-European’. - Note that the particles, on the
assumption of an OV type for PIE, are postpositive with nouns, and prepositive with verbs, insofar
as they have relational value. Cp. Waanders (1994).
2 E.g. *h,e-s(e)- ‘tobe’?, *<fie-h/e)- 'to put down’?, *de-k(e)- ‘here you are’, ‘take’?
270 F.M J. WAANDERS

going to analyze the first element of άμ-φί etc.—just let me say that there are several
possibilities. As for Sanskrit a b h i , we cannot be certain that its first syllable goes
back to * h j $ - (like Greek άμ- and Germanic i o n ·), rather than * h , e /o - , which will be
dealt with in section 7.
As for the ending -φι in M ycenaean Greek, there is no sufficient reason to
assume ablatival value for it; however, this is not the occasion to discuss this matter.3

2. *pe/o ‘aw ay a n d dow n’?

The etymology o f Latin o b is ambiguous: it may represent * h , o + * b hi or * h , o +

* p i, in the latter case with lenition o f * p when it had become final. Oscan d p has IE
* p , but 1 am not sure that this fact is decisive for the etym ology o f the Latin form,
which might even continue both * h . o f ^ i and * h ,o p i. As I stated in section 1, it is just
possible that Sanskrit a b h i continues * h , e l o - b hi, rather than * h 2N ~ b hi. However, let
us take * h , o - p i as our starting-point for Latin o b ; in any case, the * p is certain for
o p e r i ö < * o p - w e r y ô . The particle * h , e / o still awaits its turn. The reconstructed ele­
m ent *pi, which may be further analyzed along the lines indicated for * b h i, is based
on * p e /o , possibly indicating some type o f displacement, ‘down (upon)’ o r ‘aw ay’:
cp. Latin ρ δ η δ < * p Ô -s [ i] n Ô T lay dow n’4; perhaps— I say this with extreme cau­
tion—also the second syllable of Greek ά-«ό (or, with an extra element - u , ά-πύ), and
ύ-πό (*/Aj-/«-po?), with their cognates in sister-languages o f Greek. With i, we find
- p i in Greek έ-π ί, Mycenaean preposition ό-ici beside έ-π ί, and όπι- as first element
o f όπίσω, οπισθε(ν), etc., and probably also in πι-έζω (although this has been seri­
ously doubted). By the second millennium BC, or earlier, * h , e /o p - may have become
lexicalized as a nominal stem; as Morpurgo Davies remarks in a paper on Mycenaean
o - p i , e - p i, where she adduces Anatolian evidence, O ld Hittite shows clearly that we
are dealing with a nominal root with different case endings’ (1983: 306). I want to
stress that έπί/όπ1 can all the same have the o r ig in described above: the later situation
does not exclude it. A similar line o f thought could be followed concerning, e.g.,
άντί, by the second millennium BC clearly the loc. sg. o f a nominal stem
witness the Anatolian forms, but p e r h a p s originally a cluster o f particles. — It is hard
to tell whether * p e in Latin q u ip - p e , Greek τίπτε, both < * k wi d p e , also belongs here;
I have not been able to find a semantic justification for such an assumption, so I
simply won’t make it.

3 The issue is dealt with in extenso in Waanders (1997: part II).

4 Admittedly, Latin *po- may continue but this assumption is not compelling.
 PARTICULARS OF SOME PROTO-INDO-EUROPEAN PARTICLES 271

3. hje/o ‘up and away’?5*

A separative (?, ‘away-from-me’?) particle *h2elo may perhaps be assumed for the
first syllable of Greek ά-πό, Latin ab, and possibly also for that of Greek ά-νά,
Gothic ana, English on, etc. Words for ‘other’ are probably also based on this parti­
cle *h2e/o: combination with the ‘awayness’-particle *le!o® gives *hiel-, which is
found in Greek άλ-λος (< *-l-y-), Latin al-ius, and al-ter, whereas combination with
*nelo provides for *h2en~, as found in Sanskrit an-yàh, Gothic an-\>ar etc. — I
would like to add Greek α-νε-υ (and ά-ν-ις) to the list. Beekes connects ανευ with
Latin sine, Sanskrit sanutàr (1982/3: 207f.; 1995: 221). However, the assumption
that Attic borrowed άνευ from a psilotic dialect is not entirely convincing; perhaps
for that reason, Beekes brackets the initial *s. Another reason for me to reject the
connection is that I still believe in laryngeal umlaut, first proposed by Beekes, though
he himself has abandoned the idea. This means that from *snh,eu I expect teveu, or
psilotic or *r-less |ένευ, and not ανευ.
Epic-Ionic α-τερ, on the other hand, with East-Greek psilosis, and perhaps with
Aeolic barytonesis, agrees with OHG suntar, Dutch zonder, etc. It may either go
back to *sn-ter, with an original dental nasal, and be related to Latin sine (Lindeman-
form *st}t'?), Sanskrit sanutàr, or to *sn}-ter ‘on one side’, ‘apart’, deriving from
V*sem- ‘one’. In the latter case, it is not related to Latin sine, Sanskrit sanutàr, but to
Greek ατερος (analogically έτερος).

4. *kelo ‘here with me’?

One element of the PIE value of *kelo must have been ego-deixis; cp. Latin e-cce
‘look here’, ‘void’, ce-do ‘give to me’, also ci-s, ci-tra, ci-terior, and *ky in Proto-
Greek *ky-ämerom, Att. τήμερον etc. If I am not mistaken, *ke/o also expresses
‘togetherness’, perhaps originally, given its ego-ddxis, ‘here-with-me-ness’, ‘ego-
proximity’. The inner-Greek development of *ke(m) as a modal particle has been
dealt with by Ruijgh (1992); at some stage of Proto-Greek, the value of ke was ‘à ce
moment futur’. This perhaps means that the ego-deixis had been lost, but not the
‘togethemess’-value. Combative value is apparent in the forms *ko (Latin co-. Gothic
ga-, with ‘para-Vemerian’ g- [?7], perhaps also ka- in Hittite katta etc.8), *kom (Latin

5 Perhaps it is wiser to follow Melchert and Dunkel in recognizing in *h,e/o a ‘proximal’ panicle;
cp. Dunkel 1994: 33. The separative value of e.g. *k2e-p-, (and *h,e-u- ‘away’) must then be
due to the second, separative element *pelo, *lelo (and *weto, *«)-in other words: *h,e-p(o) *
‘away from (po) the proximity (h ,e)\ etc.
5 Perhaps represented in the prohibitive particle U in Hittite (*leh„ rather than a mutilated *mek,,
or *ne; cp. Melchert 1994:185). (I cannot judge the prehistory of the Hittite genitive ending -ll.)
7 Germanic ga- may have nothing to do with *ko, but derive from *gho (the only alternative, from a
phonological point of view).
®Cp. Waanders (1994:430). - Melchert (1994:126) assumes *hr}-, which in unaccented non-final
Position is supposed to yield ia - rather than koN-; this supposition (regular loss of nasalization in
said position) makes a somewhat ad hoc impression on me.
 F .M J . WAANDERS

c o m - , c o n - in com pounds, cum; possibly also to be found in Greek κ ο ιν ό ς, if <

♦κομ μ ός), and * k r n - as in Greek κα -τά ‘following’, ‘downstream ’, and κα -τι-/κ α-
σι· in Horn, κασ ί-γνητος etc. Καί, and Arcado-Cypriot κά ς, o f course also belong
here.
Note that I would not want to separate * k o ( m ) and * k e ( m ) \ I have a problem with
attributing a separative, or ablative value to the nasal o f κεν , if it is a la b i a l nasal in
origin— Lejeune assumes this, and Ruijgh with him ( 1992: 83), on account of Latin
e x - i m ‘thereafter’. Considering the comitative value of * m e / o as we encounter it in
e.g. Greek με-τά ‘in the middle o f , as against the separative/abessive value o f * n e lo ,
as I take it, I would sooner expect an original d e n t a l nasal to be added to * k e to
express ‘from that point in time onwards’, ‘dès lors’. The final nasal o f the adduced
Greek forms ένθ -εν , νυν, and κεν is not very instructive, since every final nasal in
Greek is phonemically dental (‘neutral’); it is the labial nasal o f Latin e x i m , i s t i m ,
illim which bothers me .9

5. *me/o ‘among’

The best-known G reek preposition built on * m e l o is με-τά; Gothic m i p . O ld Ice­

landic m e d , Dutch m e t , etc. probably agree in form with Greek μετά. Μ ετά already
occurs in M ycenaean, in the expression m e - t a - q e p e - i μετά κ * ε σφέΐιι ‘and am ong
them ’, and in one or two compounds. * m e + (complex) * d hi provides the basis for
the adjective * m e - d hy - o - s , which is regularly formed by means of the thematic vowel;
this adjective survives in Sanskrit m d d h y a h , Greek μέ(σ)σος, Latin m e d i u s . Gothic
m i d j i s . * m e / o is further found in words like Greek μ έ-χρ ι(ς) and ά -χρ ι, μέ-σ-φα,
μέ-σ-τε, μέ-σ-τα, for which an original meaning ‘into the middle o f is not im prob­
able. As I stated elsewhere (Waanders 1994: 430 n. 7), * m e / o may originally have
implied a plural company: this would best account for the old meaning ‘am ong’ o f
G reek μ ετά, and for the meaning o f * m e - d hy - o s , viz. ‘which is in the m iddle’.
Likewise, it accounts for Mycenaean m e - t a - k i - t i - t a μ ετα -κ τίτδ-ς ‘im m igrant’, ‘a
person who has settled among the resident population’, and Classical Greek μετα-
νάσ-της, and μ έτ-οικος .10
Μ ετα-ξύ seems to combine μ ετά and a shorter variant o f ξύν, without a final
nasal. The etym ology of ξ ύ ν is disputed. According to Ruijgh, ξ ύ ν could be
explained as the result of contamination o f * k o m and *i«;n to my mind, it could as
well be a c o m b i n a t i o n o f * k ( e lo ) and * s u . Alternatively, the initial velar may be the
reflex of the zero grade of the panicle * g he /o . Yet another possibility is to relate ξύν

9 Generally speaking, -N = -m in Latin, the notable exceptions being forms like fla m e n (m.),
in guen, neuters in -m en (all of them n-stems!). Even novem ‘nine’ < *h ,n ew $ has -m (but usually
ascribed to influence of decern < ‘ dekn)). The -m of exim etc. may also be inconclusive, after all.
10 It seems impossible to connect prohibitive μή (rm eh ,), or the first singular pronoun * (h ,)m e,
with *me ‘among’.
" The archaic particle *su (*se/o + *we/o?) later becomes the dat./Ioc. plural ending known from
Sanskrit; in Greek it has been replaced by -σι with -i after the corresponding singular ending.
 PARTICULARS OF SOME PROTO-INDO-EUROPEAN PARTICLES 273

to a putative verbal root * g he s - ‘to touch, take up, grab’, from which several ‘hand’-
words are derived: Hittite k e s s a r , Greek χειρ , Sanskrit h d s ta h . A neuter noun * g ho s u
‘touch’, ‘proxim ity’ might conceivably become a relator meaning ‘near’, ‘with’— in
the process, it lost its (stress-)accent and was reduced to zero grade. For the final
nasal o f ξύν, we can think o f diverse explanations, which I shall not sum up. M yce­
naean k u - s u , once occurring as a preposition, and a couple of times as first member
of a compound, does not inform us on the presence or absence of a final nasal.

6. *nelo ‘aw ay ’, ‘dow n’

* n e /o seems to indicate ‘awayness’, non-proximity, absence o r separation from the

immediate vicinity of interlocutors, or motion down o r away. The combination with
* y ( e lo ) , viz. * n i, with double zero-grade, is the first m em ber of ‘nest’-words: Latin
n i d u s , Sanskrit n i d d h , English n e s t , whose second m em ber is * - s d - , zero-grade of
V* s e d - ‘to sit’. * n e / o is further found in pronouns with (7/e-deixis, e.g. Greek
(έ )κ ε ΐν ο ς ( * [ h , e - ] k e - y - h , e - n o -), Icelandic k a n n < * k é - n o - s (* k e - h , e - n o - l ) } 2 It
could also be concealed in Greek έν, έν ί, Latin i n , etc., i.e., * h , e + n ( i )— thus Bader
(1981: 44f. with n. 54), whom I am now inclined to follow. The ‘inside-awayness’-
meaning o f the compound * h , e n then gave rise to derivatives like Greek έντός, Latin
in tu s , Greek έντερα, Latin in te r , in te r io r , Sanskrit a n td r , d n ta r a h , etc.
With some hesitation I would further like to suggest that the negation * n e is the
selfsame particle, stating the absence of an action, process, etc.

7. *h,e/o ‘th ere w ith ’

The (anaphoric) particle * h , e ! o , I surmise, establishes a link between entities or

actions. It is an ingredient of several pronouns, which I won’t discuss here. We
already came across * h ,e !o in connection with έπ ί, όπί. Furthermore, it is possible to
identify the augment of Greek, Sanskrit, Armenian, Phrygian with this particle. The
augment links past actions, the first past action mentioned (augmentless form) func­
tioning as point o f reference for other actions (forms w i t h the augment); * h , e then
means, in the words o f Ruijgh (1992: 83), ‘à c e moment du passé’.
W ith m uch hesitation (to m y knowledge it has not been suggested before, but
one never knows for sure) I venture to suggest that perhaps the zero-grade of the
particle under discussion came into use as the instrumental/comitative singular end­
ing, and also as the primordial ending o f the dual, i.e. nom./acc. The dual expresses
that one entity is closely linked to another one. * p o d - h , could thus be taken to mean,
elliptically, ‘foot therewith’, i.e., ‘with the other o ne'— practically speaking ‘both
feet’, ‘the tw o feet viewed as a couple’; along the same lines, ‘father therewith’ could

12 H ie e#o-de ixis o f *ke is ‘o verruled’ b y the */i,eno-elem ent; from ε κ ε ίν ο ς derives έ κ ε ΐ ‘there’
(back-formation; formally =» Lai. ecce ‘here [is, are]’).
 F .M J . WAANDERS

be a way to express ‘father and mother’, ‘both parents’, ‘the tw o parents viewed as a
couple’.

8. Conclusion

From the preceding considerations I would like to conclude with tentatively positing,
broadly speaking, as PIE ‘associative’ particles: * b he l o , * k e l o , * m e lo , * h , e io , and as
‘dissociative’ particles: * fh e lo (?), * p e ! o , * n e lo . Further investigation might reveal
finer distinctions than I have attem pted in this paper— or prove one o r more o f my
reconstructions untenable.

References

Bader, F.
1981 ‘Anaphoriques du type viv en hittite’. In: Y.L. Arbeitman, A.R. Bomhard
(eds), B o n o H o m i n i d o n u m . Essays in Historical Linguistics, in M emory of
J. Alexander Kerns. Amsterdam/Philadelphia: John Benjamins, 31-45
Beekes, R.S.P.
1982/3 O n Laryngeals and Pronouns’. Z V S 96, 200-232
1995 C o m p a r a t i v e I n d o - E u r o p e a n L i n g u i s t i c s . A n I n t r o d u c t i o n . A m ster­
dam/Philadelphia: John Benjamins
Dunkel, G.E.
1994 ‘The IE Directive’, fn: G.E. Dunkel a.o. (Hrsg.), F r ü h · , M i t t e l - , S p ä t i n ­
d o g e r m a n i s c h . Akten d e r IX. F achtagung der Indogerm anischen
Gesellschaft vom 5. bis 9. Oktober 1992 in Zürich. Wiesbaden: Ludwig
Reichert, 17-36
Melchert, H.C.
1994 A n a to lia n H is to r ic a l P h o n o lo g y (Amsterdam/Atlanta, Rodopi)
Morpurgo Davies, A.
1983 ‘M ycenaean and Greek Prepositions: o-pi, e-pi etc.’. In: A. Heubeck, G.
Neumann (eds), R e s M y c e n a e a e . Akten des VII. Internationalen Mykeno-
logischen C o l l o q u i u m s i n Nürnberg vom 6.-10. April 1981. Göttingen:
Vandenhoeck & Ruprecht, 287-310
Ruijgh, C.J.
1992 ‘L ’emploi le plus ancien et les emplois plus récents de la particule κ ε /ά ν ’.
In: F. Létoublon (éd.), L a l a n g u e e t le s te x t e s e n g r e c a n c ie n . Actes du
colloque Pierre Chantraine (Grenoble - 5-8 septembre 1989) Amsterdam:
Gieben, 75-88
Waanders, F.M.J.
1994 ‘The Genesis of Adposiüons’. In: G.E. Dunkel a.o. (eds) 1994,427-432
1997 S t u d i e s i n L o c a l C a s e R e la tio n s in M y c e n a e a n G r e e k . Amsterdam: Gieben
 G EN ER A L INDEX

aber 87,92,137
accentuation, problems of 158
additive value of καί 67
advert) 233
adversatif/adversative 135,136,142,224
adversative particles 102
answer-questions 160
αξίωμα 241
appaile nance dialectale de κε(ν) et de άν 255
Arcadia and Cyprus 52
argumentative contexts 224
Aristophane, Tkesmophorìes, particules dans 26
Aristote, particules dans 46
assent 217
assent seeking questions 157,164
asyndète, asyndeton 43,64,66
attitudinal particle 213
augment 273
‘balancing adversative’ 189
basic value of οΰν and ούκοΰν 163
borrowing 49,50,67
central discourse units 102
changement de thème 84
classements d'cmplois 80
closed class 49
cognitive unit 56
cohesion 62
conglomerates of primitive particles 269
conjunctions 234
contrastive value of δέ 67
coordinateur binaire 79
correction de posé 86
correction de présupposé 85
Cypriot 65
de dicto 104,109,116,120,125
de re 104, 107,115, 119,123
déflnition de la coordination par άλλα 97
denial of expectation 111, 114,115,118,120,125
development in the use of particles 49
development unit 56
dialogic contexts 214
276 INDICES

dialogue 76
Discourse Topic 104,110, 111, 116,120,123,124
Discourse Topic shift 103 (see also Topic shift)
distinctions sémantico-pragmatiques 80
échelles argumentatives 141
éliminatif 142
éliminatif inverse 92
embedded narrative 173
embedded sequence 103,103, 108
embedding sequence 101,103,108
emphasis, emphatic 209,211
epanalepsis 175
epic regression 177f.
euphonic 238
explanatory use of γάρ 175
fréquence des particules 21
à la scène 36
dans les oeuvres non scéniques 36
functional meaning 211
grammatical shifts 62
grammatical status o f particles 49
gcammaticaiization 50
grec écrit 17
grec parlé 17,43
grec parlé et particules 19
Greek, allegedly rich in particles 49
hiatus 238,242,243
hierarchical structure 102
importance quantitative des particules 17
interactional level 211,213,219,226
interconnected questions 161
inverted denial of expectation 123
mats 89,137
Maxim of Relevance 170
Ménandre, particules dans 44
μερισμός 233,245
δμοιον άξιώ μαιι 241
open class
parenthèses 140
panes orationis 233
particles in inscriptions 50
particules au théâtre 4 1
- dans les parties non dialoguées 26
particules dans les textes versifiés 33
 GENERAL INDEX 277

particules, frequence des · dans Ia langue parlée41

panies dialoguées 30
parties dialoguées - non dialoguées 23
pero 137
Platon, Apologie de Socrate, particules dans 21
Ménon 21
Protagoras 23
poetry 49
POP 101
posé 81,95
pragmatic, pragmatically, pragmatique 75,135,191,211
pre-sequence 104,108,11 If,, 116,118
presentational level 191,211
présupposé81,140
progressive 227
progressive value of μην 226
prose 49
prose inscriptions 52
προσθήκη 239, 240
punctuation conventions 161
PUSH 101,184
referential meaning 211
representational level 191,211,219,222
rhème 82
rupture 141
rupture discursive 91
rupture discursive contradictoire 93
sentence connective 62
sino 137
Solon, particules dans 33
sondern 8 7,92,137
Sophocle, Oedipe à Cobne, particules dans 28
sous-entendu 81
sous-entendu virtuel 90
στοιβή 238,240
Stoics 236
style, stylistic 234ff., 239f., 244
subjonctif avec κε(ν) dans la principale 257
subjonctifs en -μι, -σθα et -σι 254
subsidiary discourse units 102
syllogisme 135, 137,144f.
text structure 102
textes scéniques - non scéniques 31
thematic shift 62,65
278 INDICES

ihèmc de discours 81
Topic shift 197,198,203 (see also Discourse Topic shift)
topique 82
transitive function 65
truth of a statement 215,218,220
unité sémantique 76,77
variety in the use o f particles 49
verse inscriptions 52
χασμώδες 239
Xénophon, Banquet, particules dans 25,47
yes/no questions 157,166
άλλα 25,26,28,29,30,36, δή 25f„ 28f-, 31, 36f., 39,
37,39,42,51.52,53, 42, 167, 172, 211,216,
77ff., 105,106,110, 111, 239
116,113, 141,145
άλλά γάρ 113ff. δήπου 36, 37, 39
αλλά μήν 144 δητα 28. 29, 31, 36, 37, 39
άλλ' ού 92 ε ις ο 250
κε(ν)
δμ-φω 269 ε ίτε 51
άν 248,251 έπήν 256
ά-νά 271 έ-πί 270
δ-νε-υ 271 εύχε 51
ά-πό 270,271 Ά 25, 26, 28, 29, 31, 36,
39, 42, 52
δρα 2 3,25,31,36,37,39, ί 29, 31, 37, 39, 51, 210,
42,162,167 218
άρα 23, 25, 29, 31, 36, 37, ή δρα 221
162, 164,166
δρ' ούν 164, 166 ή γάρ 151, 153
άχάρ 53 ή δή 221
αΰ 25, 26, 31, 36, 37, 39. ή μήν 218
42
αύθε 51 ή που 221
αύτάρ 51 ή τάρα 221
γάρ 2 3 ,2 5 ,2 6 , 28, 29,31, ήδέ 51
36, 37, 39, 42, 51, 101,
110, 113, 184
γε 23, 25, 26, 28, 29, 31, ήν 256
36, 37, 39,42, 211
γ ε μήν 225, 244 23,25, 26, 28, 29,31.
36, 37, 39, 42, 43,51,
57,64
γοΰν 36, 38, 39 καί ... αΰ 195
δέ 25, 26, 28, 29, 31, 36,
37, 39, 42, 51, 52, 55,
56, 5 7 ,5 8 ,6 0 , 62ff„
162, 187, 189, 191
δέ γε 144f.
280 INDICES

καί ... 179 μηδέ 29, 37, 39

γάρ
και ... γε 195 μήν 23 , 26, 31, 36, 37 , 39,
42, 209, 211,218, 223
και (...) 187ff. μήτε 29, 36, 37, 39
δέ
κ α ί ... δέ 206 νεμεσση- 252
ωσαύτως τόν δέ
κεν ειη
και δή 216 νυ 250
κ α ί ... δή 195 νυν 28, 36, 39
και μήν 2I5ff., 228,244 δττι κεν 253
είπω
καί νύ 249, 260 ού 210
κε(ν)
καίτοι 37, 89, 105, 122ff., oi>yàp 78
1331Ϊ. άλλά
κάς 52, 66, 250 ού κεν 256
κάς/καί 52, 55, 65 ού μέντοι 78
άλλά
κείνος 273 ού μήν 22378
κε(ν) 248,251,255,2581. ού μήν 78
άλλά
κεν πολύ 252 ούδέ 25, 26, 28,29, 31,36.
κέρδιον 37, 39, 42
εΐη
κοινός 272 ούδέ ... 187, 204
δέ
μέν 25, 26, 28, 29, 31, 36, ούκ δν 256
37, 39, 42, 51. 52, 55,
211
μέν ... δέ 68 ούκ ούν 163
μέν ... 205 ούκουν 158f„ 164
καί ... δέ
μέντοι 2 5 ,2 6 ,3 1 ,3 6 , 37, 39, ούκοΰν 23, 25, 26, 31, 36, 37,
42, 105, 114, 116, 122 39,42, 145, 158, 162
μέ(σ)σος 272 ούν 25.26, 28, 2 9 ,3 1 ,3 6 ,
37, 39, 42, 101, 162,
165, 167
με-τά 272 ούνεκα 178
μετα-ξύ 272 περ 51
 281

πότερον 162
... ή
πω 51
τε 25, 26, 28, 2 9 ,3 1 , 36,
37, 39, 42, 51, 52, 53,
65
τε ... καί 51,196
τήμερον 271
τ ίδ έ 169, 170
τ ί δη 172
τί οΰν 168, 170, 171
τί οΰν δή 173
ποτέ
τί οΰν δή 171
τοι 29, 36, 37, 39
τοίνυν 23, 26, 31, 3 6 , 37, 39,
42
ύ-πό 270
-φι 269
 INDEX O F PASSAGES DISC U SSED

Aeschylus Λα. 482 85

Λα. 497-98 79
Pe. 406-7 227 Λα. 527 94
Pr. 771-3 220 Λα. 598 91
Pr. 869-73 225 Λα. 646 84
Pr. 981-2 218 Λα. 788 95
Pr. 377-80 160 Λα. 946 93
Pr. 757 151 Λα. 997 89
77i. 529-32 223 Λα. 1025 91
Λα. 1039 88
Apollonius Dysc. Λα.1043 92
Λα. 1066 94
pror. 50.11 238 Λα. 1130 85
Λα. 1180 78
Apollonius Rhod. Λα. 1199 94
Λα. 1298 87
Argon. 1.1-7 183

Aristophanes Aristotle and

[Aristot.]
Eq. 810-11 160
£<7.820-1 160 Probi. 919a 234
Nu. 1462 189 Rhet. 1407aI9ff. 236
Pax 243ff. 201
t o . 23-8 161 [Demetr.]
£<2.45 86
Λα. 56 87 Eloc. 55ff. 239
Λα. 57-58 79
Λα. 103 95 Demosthenes and
Λα. 123 84 [Dem.]
Λα. 164 201
Λα. 174 95 21.189 192
Λα. 202 94 42.1 194
Ra. 226 83 57.4 193
Ra. 2 Π 78 59.126 205
Ra. 258 86
Λα. 292 94 Diogenes Laert.
Λα. 439 87
Λα. 469 88 7.67 236
 INDEX OF PASSAGES 283

Dionysius Halic. 3.71.3 118

3.80.4 124
Dem. 39.212.20ff. 241 4.32 107
Dem. 40.215.19ff. 242 7.17.2 112
hoc. 3.58.18ff. 243 7.101.2-3 119
7.104.1-2 126
Dionysius T hrax 7.143.1-2 113
7.148.4 124
86.3f. 238 7.158.1-3 110
7.159 111
Euripides 8.3.2-4.1 109
8.71.1 123
Ale. 384-6 228 8.79.4 111
Ale. 642 221 9.71.2-4 125
Ale. 658-9 224 9.82.3-83.1 121
Ale. 1117-8 217 9.109.3 112
Ba. 807-8 216
Ba. 1043-5 181 Homer
El. 239 160
El. 907-12 224 II. 1.8-16 177
Hel. 784 152 11.2.11-13 176
1A 325 152 II. 2.317-9 176
Med. 1116-9 229 II. 6.37-43 178
Ph. 697-700 220 Od. 4.5-7 176
Rh. 284 225
Lysias
Herodotus
1.5-6 182
1.14.4-15.1 103
1.36.3 117 P indar
1.59.1 175
1.78.2-3 120 O. 7.24-9 179
1.87.3-4 109 P. 4.68-70 180
1.103.3 -104.2 116
1.108.4-5 104 Plato a nd (Plato)
1.117.4 116
1.130.1-2 120 Ale. / 110e4-6 141
2.44.1 200 Ale. 1 106a3 141
3.16.2-5 121 Ale. II 147b8 190
3.32.3-4 117 Ap. 17a 1-4 136
3.69.4 115 Ap. 27a6 142
CArm.l59d3-ll 144
CArm.l64blI-c6 144 Phil. 13b 192
Cra. 40Iel-3 134,139,146 Pn. 320c 183
Euthydem. 300b 156 Pn. 331c 154
Eulhyphro 3bl0-c7 142 Resp. 440d4-6 134
Grg. 448a6-10 171 Resp. 464a-b 146
Grg. 449a2-b8 165 Resp. 583b3-7 134
Grg. 449el-450a5 165 Resp. 607c 155
Grg. 450b3-5 173 Thl. 148b5-8 135
Grg. 450d 195 Tht. 160e 155
Grg. 452d8-e5 134 Tht. 171e 193
Grg. 453b4 172 Thl. 202b 189
Grg. 457el-3 172
Grg. 459b6-c5 159 Sophocles
Grg. 462c4-d2 163
Grg. 463c6-d5 169 Aj. 850 222
Grg. 466d7-e3 159 Ant. 155-6 228
Grg. 469al-5 172 Ant. 526 228
Grg. 469bl0 153 Ant. 557-8 218
Grg. 474bl-c3 159 Ant. 626 228
Grg. 475a 199 Ant. 1053-4 225
Grg. 47ft2-477a3 162 Ant. 1180 228
Grg. 477a5-b8 166 Ant. 1257 228
Grg. 478a8-bl0 164 El. 385 221
Grg. 482b 137 El. 554-6 215
Grg. 494e 154 El. 554-8 217
Grg. 497d6-498b4 170 El. 968-74 226
Grg. 503bl-c3 171 El. 1221-1222 152
Grg. 506d 195 El. 1349-53 219
Grg. 5ilb3-6 160 OC 26-8 214
Hp. Mi. 363c 155 OC 81445 222
Ion 533c 148 OT 290 227
La. 181b 199 ΟΓ810 223
Lg. 833a-b 194 OT 1039 151
Men. 77c 154 OT 1237-42 181
Men. 87c 195 Ph. 809-11 213
Phd. 68d9-12 138 Ph. 1257-8 147
Phd. 68dl0-69a2 143 Ph. 13601Ï. 198
Phd. 68dll-12 133
Phd. 77a6-10 138 Thucydides
Phdr. 241d4 139
Phdr. 264c3-7 140 1.132.4 204
 INDEX OF PASSAGES 285

1.144.2 190
2.36.1 189
6.71.2 197

Xenophon

Ages. 10.1 200

An. 1.8.2 192
An. 5.3.9 194
An. 5.6.15 194
An. 7.1.30 205
Cyr. 1.4.26 202
Cyr. 3.3.19 198
Cyr. 7.1.1 203
Cyr. 7.2.10 202
Mem. 1.1.15 194
Oec. 1.3 203
Oec. 4.20-24 153