‘THE GREEK PARTICLES BY J. D. DENNISTON FELLOW OF HERTFORD COLLEGE, OXFORD UNIVERSITY LECTURER IN GREEK AND LATIN LITERATURE SECOND EDITION vesOsford University Press, Amen House, London E.C.4 GLASOOW EW YORK TOROHTO MELBOURNE WELLINGTON BOMBAY CALCUTTA MADEAS KARACH! CATE TOWN TBADAM Geoffrey Cumberlege, Publisher to the University NESE Poa UNF De' Te TIS _ miasr apiTion 1934 SECOND EDITION REPRINTED LITHOGRAPHICALLY IM CREAT BRITAIN AT THE UNIVERSITY PRESS, OXFORD, mow COMBECTAD SHEETS OF THE FIRST EDITION 1954 PREFACE TO THE SECOND EDITION DuRING the fifteen years between the first edition of this book and his death in 1949 Denniston made notes of a large number of additional examples and on many points changed his mind in the light of this fresh material. My principal task in the prepara- tion of this second edition has been to incorporate all these addi- tions and corrections. For the sake of speed and economy photographic reproduction from the first edition has been em- ployed. This has meant that no insertion could be made in the text without an omission of corresponding size on the same or the next page. Accordingly, I have reduced many quotations to bare.re- ferences; the choice of what to omit or condense has not always been easy, but in all cases I have weighed the new matter against the old and omitted whatever contributed least to the argument. Above all, I have taken the opportunity to redistribute matter between the text and the addenda in such a way that, as far as possible, the addenda do not accumulate additional examples but are confined to discussion of difficulties of text and interpretation. The notes which Denniston made in his interleaved copy nearly always made it quite clear what he wanted to insert and where; some other notes, in the margins of review offprints and correspon- dence, did not make it clear, and I have accordingly been very cautious in using them. In general, I have tried to admit nothing into the text without being certain that it represented Denniston’s considered view. But in a few cases the fresh examples which he had added seemed to me to necessitate a slight modification of his original views, and I have rewritten a sentence or two accordingly (p. 188, on postponed 8€ in Middle and New Comedy ; pp. 290-1, on xaé linking qualitative attributes; p. 462, on 8° ov in the sense of d& 84; p. 501, on te linking qualitative attributes). As thoroughgoing a correction as possible has been made of the few printers’ errors and fewer wrong references which appeared in the first edition. Inevitably, in carrying out a revision of this kind one is faced with two temptations ; to add material of one’s own, and to modity interpretations of the author's with which one disagrees, Thex FROM AID TO THE READER (Bex ST EDITION) I use italics in translation to mark the word stressed in my English: this is not necessarily the word rendering the word stressed in the Greek, On the whole, | have avoided the indication ‘etc.’, which in a work of this kind is often dangerously ambiguous. ‘/d. saep.’ denotes that the particle or usage occurs often in the author last cited, ‘#5. saep,’ that it occurs often in the work last cited, ‘e¢ saep.’ that it occurs often in Greek as a whole. Where a particle is given in brackets as an emendation, it is to be taken as a substitute for the particle under discussion. E.g. on p. 470 (s.v. 66) ‘(yp Reiske)’ means that Reiske conjectures ydp for 6. But I have been more explicit in cases where ambiguity was to be feared. AID TO THE READER (SECOND EDITION) References to Bacchylides are to the edition of Snell’ (1949). Fragments of lyric and elegiac poets are numbered as in Diehl's Anthologia lyrica Graeca {second edition); tragic fragments as in Nauck; comic fragments as in Kock; but Arn., Diehl, Mette, and Mette (Nachtrag) after a tragic fragment refer respectively to the Supplementum Euripidenm of von Arnim, the Supplementum Sophocteum of Diehl, the Supplementum Aeschyleum of Mette, and Nacktrag eu dem Supplementum Aeschyleum; Dem, after a comic fragment refers to the Supplementum Comicum of Demiaii- czuk. Fragments of Pindar are numbered as in Bowra, with Schroeder's number in brackets; of the Presocratics, as in Diels (fifth edition); of the historians, as in Jacoby; of Epicharmus and Sophron, as in Kaibel’s Comoediae Graccae Fragmenta. An asterisk indicates that the Addttzenal Noles at the end of the book should be referred to. ee ee eee CONTENTS? FROM AID TO THE READER (FIRST EDITION) ix - AID TO THE READER (SECOND EDITION) . x INTRODUCTION. . 2 evil 1. THE ORIGINS AND FUNCTIONS OF PARTICLES — xavi (1) Definition of ‘particle’. “Particles originally other forms of speech. saxvii (2) Particles denoting a mode of thought in isolation. Emphatic Particles: affirmative, intensive, determinative, timitative xxxvii (3) Particles conveying moods of emotion, nuances =. xxviii (a) Particles of emphasis and nuance grouped as ‘adverbial’ Difficulty of rendering these particles : toxic (5) Particles establishing a relationship between ideas. Con- nective, hypotactic, ‘responsive’, apodotic and resumptve, and ‘corresponsive* uses. : : Doo Il, CONNECTING PARTICLES . : : : «xiii (1) The origin of connectives : : : : + xiii (2) Connexion and asyndeton. Cases where asyndeton is often employed. Polysyndeton. Use of asyndeton for emotional effect : + xdiit (3) Nature of connexion omitted when asyndeton is employed | xlvi (4) Tests of admissibility of asyndeton ina givencase . 0. xlvi (5) Apparently superfluous connexion. (i) In answer to question or command. (1i) At opening of speech or work. (iii) At opening of reported speech. ~ xbi (6) The different methods of connexion : additional, adversative confirmatory, and inferential. Distinction between ad tional and ‘progressive’ uses. *Eliminative’ and * balancing adversatives . . . . = xlvti (7) Abnormalities of reference inconnexion =. =. wh II. COMBINATIONS AND COLLOCATIONS OF PAR- TICLES. Ki (1) The distinction between combinations and coltocations. (i) Change of meaning in combination, (11) Dependence ot one particle on another. (iii) Other tests of coherence. {iv) Fortuitous collocations, “Gravitation of certain particles towards certain other parts of speech . : (2) Avoided collocations —. : : . : . li (3) Split combinations (4) Exceptional combinations (5) Double connexions Vin the summary of the test square brackets denote an unimportant, iNlusary, of highly doabttul usage.xii CONTENTS@ 1V. DIVERSITY IN THE USAGES AND MEANINGS OF PARTICLES. Deviations from normal meaning even in case of apparently stereotyped idioms. Ambiguities more frequent in case of combinations. Occasional logical separa tion of two particles which normally form a combination. Different meanings in close proximity “wi V. THE POSITION OF PARTICLES . . : (1) The position of particles in sentence and clause. Adverbial particles gravitate to opening. Position of enclitics. Posi- tion of ye, 8, and adverbial cai relative to emphasized word. Position of connectives. Types of postponement : (2) Order of precedence in combinations. (i) Adverbial particles and connectives. (ii) Preparatory particles and connectives. (iii) Preparatory and adverbial particles. (iv) Two adverbial particles : : : : : : : kk Wii Wii VI. THE STYLISTIC IMPORTANCE OF PARTICLES ki (1) Repetition of particles. Greek tolerance of repetition. signed repetition. ‘Gregarious’ tendency of particles xii (2) The employment of particles in different periods, dialects. and styles, and by different authors. Difficulty of inquiry . (3) Chronological differences. Exacter delimitation of functions in post-Homeric Greek. Emergence of stereotyped com- binations. In other cases increased diversity of usage. The development of 89, ogy, uv, and other particles: obsolescence of others. Development in individual authors: Aeschylus, Plato. Post-classical usages in Hippocrates . : ‘ {4) Differences in dialect. Tonic usages in certain Attic writers. ody and roe combinations. Other examples. Tonicisms in Sophocles. : : : : : + xx (5) Differences in genre. Dialogue and continuous speech. Par- ticles especially common at opening of answer: usually omitted in oratro obliga, but sometimes retained. Usages transferred from answers to continuous speech (imaginary dialogue). Certain particles mainly confined in Homer and the historians to speeches. ‘The orators stand midway between dialogue works and formal treatises. Vivid usages in Demosthenes. Political and forensic oratory (6) Colloquial and poetical uses. Possibly colloquial uses in Homer. Colloquialisms in Euripides, and occasionally in Aeschylus. Epic uses in Aeschylus, Other uses peculiar to the high style, Certain Epic particles not found in later Greek. Differences between verse (including comedy) and prose. Ixxv Isiv bv O)! ‘al preferences of various authors, Usages of Demos- vies and pseudo-Demosthenes compared . Isxviii "AMG . : : . . : : . ot 1, Genewat Apvixsative Use . . . cn {4) Eliminative. : . t fiy Usually either(apddda clause or (4) clause to which it is opporedd co CONTENTS xiii q Gii) Particular varieties of dda following negative clause. (a) ob psvov (obx Sas)... GAAG cai, etc. (6) ddAd, following nega~ tive clause, meaning ‘except’. Xj» ddd.” (c) With com- parative adverb in negative clause. (d) ydddd. (e) Rhetorical ¢ question taking place of negative clause : 3 jc (2) Balancing. (i) Without preceding yi. (ii) wiv... ddNG (@AN Syws). ddAd answering negative pév clause . Ss (3) GdAd expressing opposition in general. (i) In answers. (ii) In continuous speech. (iii) In anticipated objections (usually ANA vH Bia) : : : . : - 7 * Il. SPecIAL Uses . . . . . . . 9 (1) Following a rejected suggestion. (i) Introducing a question, ‘Well, what?’ (etc). (ii) Alternative suggestion offered. (iii) 3° aad. (iv) Hypophora : : : 9 (2) In apodosis (often adda... ye: for Add’ ody, see p. 444). .ou (3) With protasis understood, ‘ At least? : : (4) In commands and exhortations. Sometimes repeated at short interval . . . : : _ oe : (5) Inwishesand prayers. (i) Answer taking form of wish or prayer. (ii) Wish or prayer expressed during course of speech (6) Assentient. (i) Practical consent, With echoed word, especially in Plato, Consent implied : (a) enjoined task described as easy or unodjectionable ; (8) person conveys his readiness to speak by speaking. (ii) Assent, expressed (2) by favourable judgement of preceding words : (4) by form of words implying that what has been said is correct. (iii) Expressing (a) ac- quiescence: (4) a sympathetic reaction : : (7) Introducing substantiation of hypothesis or wish . : : (8) Inceptive. (i) Adversative. (ji) Response to invitation to speak. (ili) Response or approval in general . : ; i (9) Progressive. Sometimes adda xai, d\n’ of8é ene IIL, Position. Postponement after apostrophe or oath 1V. Compinations : . : (1) aad ye. For dAAa... ye, see 112 . : . : (2) aan’ of86, Why, noteven . ..’. For sense ‘Nor, again’, see Il. 9 (3) ddd’ §. (i) Negation, containing word of comparison, followed by exception. (ii) Negation, not containing word of com parison, followed by exception, (iit) Instead of general negation, particular instance of it given, Explanation of the combination dN : . . . : (4) aX? §. Usually at opening of answer: sometimes following exclamation or apostrophe. Only used in questions. iXX’. J hardly a distinctive usage. GAN’ f wrongly read for JAN F. (5) ob piv dddd: ob piv... ddAd. (i) General adversative use. piv... 00 wiv GAG. (ti) Introducing supplementary argu- ment taking marked precedence over previous one. (ut) Argument thus stressed representing second line of defence. od pi dda i . . (6) ob perros add«i (7) od yap adda 3 2 e‘Arap : xiv CONTENTS ‘Apa (ap, pa) - . : : : I. PRIMARY UsE, EXPRESSING LIVELY FEELING OF INTEREST II, SECONDARY USE, EXPRESSING SURPRISE ATTENDANT UPON DISHLLUSIONMENT (1) Verb in present. (2) Verb in past. diner» etc. (ii) with imperte in future. (4) adr’ dpa Special uses: (i) with ct, especially of ei. (3) Verb II]. PARTICULAR VARIETIES OF SECONDARY USE (t) @ dpa : : : . . : : : (2) In reported speech, and after verbs of thinking and seeming. Especially os dpa : : . : . {3) In questions, following an interrogative. (i) Direct questions. Gi) Indirect questions. aris apa : : (4) Logical IV. Position. Usually second word when connective : but postpone- ment not infrequent : V. COMBINATIONS. GAA’ dpa, GAA"... dpa. etre dpa. ofre dpa. abrap pa. yt dpa, dpa ye. 3° Spa. 84 dpa. xai pa. pev dpa, obv Spa. +’ dpa. BoBai dpa. For } dpa, da, see p. 284. For vip dpa, yép pa, see p. 56. For row dpa, dpa ro, see PP. 554-5 : : : : : : Mpa L. EQUIVALENT IN SENSE TO dpa. (1) Adding liveliness. (2) Mark- ing realization of truth or drawing conclusion, (3) «! dpa. (4) 4s dpa. (5) Following interrogative pronoun : II. As INTERROGATIVE PARTICLE. (1) Leaving answer open, (2) Ex- pecting negative answer. (3) Expecting positive answer. (4) Sp" ot. (5) dpa pi . . . . . II. Posttion. In I normally second or third word : in New Comedy sometimes first word. In II normally first word, but often ostponed by dramatists and Plato. Repeated interrogative ipa. Occasionally introducing indirect question IV. COMBINATIONS. (1) dpa ye, Apa... ye. (2) GAA’ dpa. (3) dpa 89. (4) &p ofv, (5) nai dpa : : : . : (1) Adversative. (2) Progressive. (3) wév... drip. (i) Strong ad- versative force. (ii) Weaker adversative force. (4) Position, Occasionally postponed after apostrophe. (5) Combinations. Grip Bq, drap ubv. drip oby 9. dedp re. drap ro. For érap sui, dvap ot8e, see p. $3, for drap wév, p. 391 « Abrdp (1) Strongly adversative. (2) Weakly adversative or purely pro- gressive. (3) Apodotic . . . . . 42 44 46 48 50 gt st 55 35 t CONTENTS, xv Tdp . . . . . 56 1. CONFIRMATORY AND CAUSAL. 38 IL, EXPLANATORY. (1) After rexpipioy 8¢, onpeiov &é, ete. (2 an expression denoting the giving or receiving of en Ate (3) After a forward-pointing pronominal adjective or adverb, (4) After a neuter superlative adjective. (5) Miscellaneous 58 II. PECULIARITIES IN THE USE OF CAUSAL AND EXPLANATORY yp 60 (1) Giving the motive for saying that which has just been said. 60 (2) Connexion of thought sometimes lacking in logical precision . 61 (3) Meaning ‘for otherwise 62 (4) Referring to remoter context : : : 63 (5) Used by a speaker in continuing his own train of thought 63 (6) Successive yap's with a common reference : . . 64 (7) yép referring (i} to single clause : (ii) to individual word or phrase 65, (8) Introducing an instance : : : : + 6 (9) Appositional, ‘that is to say’. (i) With pév and 36. (i) With disjunctive j. (iii) Other cases. For appositional xai yap, see pe 109 (3) . . . . . . 67 IV, Anticipatory. : : : : : - & (1) Parenthetical. With vocative preceding yip clause. In sentence opening with «ai : . - 68 (2) Anticipatory in strict sense . : : : + 69 (3) Resumption or inception of main clause marked by particle or demonstrative pronoun. (i) 8. (ii) d\Nd, hortative or ad- versative. (iti) 1. (iv) voy. (v) oy (oby 35), (vid 84 dh ody). (vii) Sue ratro, Sed 4%. (viii) ue (doubttul), (ix) Demonstrative pronoun : : : + 70 (4) Fusion of clauses. . . V. IN ANSWERS. (1) Supporting implied assent : (2) Supporting implied dissent. : . . (3) Providing motive for language used by previous speaker (4) Presupposing qualified agreement. . (5) Connexion of thought obscured by compression . > . (6) Answer in form of question. (i) Rhetorical question giving grounds for implied assent. For elliptical ros yip of ; see p- 86. (ii) Surprised question, throwing doubt on grounds ot previous speaker's words, (ui) Asking why cause of w*: precedes has been brought into operation, iv) Conve . surprised recognition of grounds of preceding question (v) Implying that preceding question need not have been put: ‘Why, of course : : : . (7) Statement or question referring to subordinate clause or indis word. : : - oe Loos (8) Exclamation, apostrophe, or question interposed before yap clause Explanation of yp in answers : Looe yes, for’, ‘no, for’. ava bxvi CONTENTS (% VI. ProcRessIVE USE, IN ANSWek-QuEsTIONS (1) Speaker proffers new suggestion afer elimination of previous hypothesis . (2) Speaker, after being satisfied on one subject, wishes to learn something further. (i) Further information required con- cerning cause of facts already known. (ii) Non-explanatary Supplementary information required. (i) and (i) with elipse: vi yap 5 « . Progressive yp in questions in continuous speech, Explanation of progressive yip in questions : VH. EL.ipricat Questions: ri yip; (cf. pp. 82: P- 285): ob ydp ; moe yp; mae yap of; VIE Assentient. (1) In general. (2) With word echoed from preceding speech. (3) yap rot (for which see, in general, Pp. $49-50), conveying assent, and adding something to it, (4) yee expressing approval. 5) Assent or approval, with 4 rdp5 (see echoed from preceding speech . ss IX. IN WISHES, ¢! yap, al ydp . . . . . (2) The wse of. (6) Conditions, (i) Wish-conditions, (ii) Pure wish-clauses. . . . (2) The use of pip. (i) Possible causal relationship. (ii) In dialogue, ‘expressing wish that something stated or wished by previous speaker might come true, (ii) Wish for something supple- mentary to fact just stated. ‘Exclamation sometimes pre- ceding wish in (i), (i), and Gil. Gv) Logical connexion not falling under above heads . : : Summary of evidence regarding the significance. of pip in ef yép wishes. Supposed use of ydp, apart from eet) rea 4) Other instances. - 333 538xxxiv Tow . . rou implies an audience. CONTENTS Examination of this principle . L. In Direct STATEMENTS (1) Im general. (2) Boasting. (3) Threatening, (4) Hortatory, deprecatory, ete, (5) In response to command, (6) Reveal- ing speaker's emotional or intellectual state. (7) Conveying criticism of previous speaker's words, (8) With gé, summons to attention. (9) Directing attention to sight ‘or sound, ‘Lo!', ‘Hark!* (10) With proverb or general ref (11) In negative statements, ofror. ofrov.. uj. covron. fro. yap efees. ofror.. oi8i.. (13) In potential with av: by crasis, rdv. For § rv, see p. 554 Il. In oTHER IsperEnpEnt CLAUSES : : : (2) Questions. (2) Commands, postive and egative (3) Prayers and wishes : . TI, In SuBoRDINATE CLAUSES : : : (1) Causal, éei rox (eat), drt roe. (2) In conditional protasis. (3) In relative clause. (4) In final clause. (5) In indirect speech. (6) wa roe... ye, with infinitive : . IV. Position Usually early in sentence, but sometimes (1) Early in apodosis, especially of conditional sentence. (2) Generally, late in sentence or clause. (3) Often between article and substan tive, or preposition and substantive. In tmesis . V. REPETITION OF ror . VI. Cosmminations. For xairot, uévror, rocyiprot, roivwy, see those particles. (1) dAXd rox, dAAd. sor. GAA" obror.. ye, GAN" oly... ro. (2) abrdp ror, drdp ro. (3) ydp rou GdAa yap ro. For xai yap to, see pp. 113-14 : : (4) pro. (i) Pro tanto reason for accepting proposition. (ii) Re- strictve 9¢ rei in general, (i) ye emphatic or exclamatory with roc standing apart. ye T0184, y¢ roi ov (5) Bé ron aa... 8 rou od8é raL, GAA obBE rot. Kad yap OBE For. (6) 84 ro, rov34. (i) 3ij rovafter relatives in Epic. Occasion- ally in prose. da 8% ro: (Plato). yap 8% ro. (ii) ror 3q. ofror By, ofru 8h... ye : : (7) Heo "As ‘either’. rot, ‘or’. (8) frog rou, Fein fripa rot piv. (ii) With im- perative or optative (9) rot dpa (répa). dpa ron. frdpa... f. dp ron ipa oi (10) »6 ros, viv vo 11) obrot piv oby Kaira. . . . . . (i) Adversative, (i) In general. (ii) Used by speaker in pulling himself up abruptly. (1ii) Objection introduced by xairos countered by a following adversative clause, (iv) Furecast- ing of following adversative by yéo, (v) Karely at opening of specch. (vi) In parenthesis. (vii) With participle 537 537 539 539 $45, 545 546 546 5a? 547 548 548 550 553 555 555 556 a CONTENTS XXXV (2) Continuative. xairor xai soe + 339 G3) Logical. (i) Rarely in complete syllogism. (ii) Usually con clusion of syllogism left to imagination. (ini) Relationship Of anos semtence to preceding sentence usvally, if regarded in isolation, adversative: but sometimes positive 560 (4) Combinations. (i) eaiton ye, xairoc... yt. (iil xairor wep (see P- 559 (vii). [xairot ye pov) . . 564 kairo in cra! . . . «$64 Tovydp, Toryapoty, Torydproe . : . . - 365 Tovyap - . . . . . + 365 (i) In Homer, used by person preparing to speak or act at another's request. (ii) In subsequent Greek, wider range of meaning. 365 5 Tovyapotv, Toryaproe . . . . . + 566 ‘Totvw - : . . : . : : - 568 Essentially an Attic and colloquial particle : 368 I. Loctcan (1) In general, (i) In continuous speech. (ii) fn dialogue, with an- swer springing from words or attitude of previous speaker . 569 (2) In conclusions of formal syllogisms (rare). (3) Responding to invitation to speak. (4) Conveying comment on previous speaker's words. Especially calor ruimy, etc., olten with ellipse of verb. (5) At opening of narration announced in advance. (6) At opening of set speech (Nenopton). (7) Rounding off long argument. + 57t IL. TRANSITIONAL . sta (1) Marking fresh stage in march ofthought, (i) Dialogue. wi Con- tinuous speech . (2) Introducing fresh item in series. (i) Dialogue. (i speech, fri roivy. roy. ..én 373 (3) Transition from enunciation of ‘general proposition to considera- tion of particular instance. Introducing minor premise. (4) General proposition implied irom particular instance of its application . : . (5) Seldom after light stop. Continuous (6) Apodotic II. Cospiations. (0) wai rointy, nai... role (2) 3) rane, [rainy di). (3) Other combinations : . SI? IV. Position ADDITIONAL NOTES. BIBLIOGRALTHY . INDEINTRODUCTION I. THE ORIGINS AND FUNCTIONS OF PARTICLES. (1) Difficult as it is to arrive at a satisfactory definition of particle, an attempt must be made at the outset. I will define it as a word expressing a mode of thought, considered either in isolation or in relation to another thought,' or a mood of emotion. It is a probable assumption that the evolution of particles repre- sents a relatively late stage in the development of expression. Their existence betokens a certain self-consciousness. A few Greek particles can be clearly seen to have been, at an earlier stage, other parts of speech. Thus ¢AAd was originally dAAq, “other things’, and ror (pretty certainly) the dative of the second person singular pronoun. mov was probably ‘somewhere’, and the tot in rorydp a case (perhaps the instrumental) of demonstrative 16. So in English ‘well’, ‘come’, ‘now’, ‘why’, have come to be used as particles. A loss of definiteness has been accompanied by increased subtlety of nuance. There is less body, more bouquet. (2) The particles which, in origin, express a mode of thought in isolation are ye, oy, %, Onv, wry, wep, rot, mov. Of these, rot presses an idea upon the attention of the person addressed ; ‘IT would have you know (or remember)’: rou conveys doubt, ‘i * This distinction cannot, however, be rigidly maintained everywhere. While in the case of adjectives and adverbs, and verbs derived irom ad- jectives, emphasis may be added without any external reference (Kady y¢, ‘ Fine!" £8 ye,‘ Excellent !": Evrvxa ye, 1 am lucky"), emphasis on sub- stantives and most verbs necessarily implies a contrast with some other thing or action, however dimly the contrasted idea may be envisaged. «It's a cloud!" (sc, ‘not a mountain top’, or ‘not anything else"), And this external reference, which underlies what 1 shall call * determinative’ emphasis, becomes patent in limitative emphasis: ofuai ye, ‘1 taint 0° (re, “but I may be mistaken’). Hence limitative ye comes near to mer in sense. Conversely, piv sufifariwmt often approximates to ye, and cai from meaning ‘even’, ‘also’, sometimes comes tw be little more than a parnele of emphasis, when the external reference which ‘even’ and ‘also’ imply is only vaguely conceived.Xxxvi INTRODUCTIO'® suppose’. The remainder primarily carry emphasis. Further, emphasis may take different forms: (i) Affirmative, denoting that something really and truly is so: (ji) Intensive, denoting that something is very much so: (iii) Determinative, concentrat- ing the attention on one idea to the exclusion of all else: (iv) Limitative, implying that beyond the prescribed limits the reverse may be true. Naturally, fixed lines cannot be drawn between these forms. Thus (i) ‘I am really sorry’ implies, almost of necessity, (ii) ‘Iam very sorry ’. (i) ‘It's really James" suggests (iii) ‘It's James and no other’, In certain contexts (ili) suggests (iv), We should not therefore expect to find, and we do not in fact find, precise delimitation of the usages of emphatic particles. Affirmation is expressed par excellence by #, which (as its regular position, first word in the sentence, indicates) affects the thought asa whole: while 4% and ye tend to cohere with the preceding word. # woAXol rodro motoior, ‘in truth many do this’: in moA- dod 8} rovro mouobae, 6 is almost an adverb, going closely with wodAof: but not quite an adverb, and moAdoi 4y, ‘really many’, is not quite the same as pada modAoi, ‘very many’. Of the other emphatic particles, zy perhaps comes nearest in force to #, though less subjective in tone: and in Homer of pofy in negation appears to be the counterpart of # in affirmation, The intensive and determinative functions are shared by ye, 4¥, and mep: limitation is expressed by ye and wep. Taking Greek as a whole, ye is the particle most commonly used for expressing determination and limitation. Interrogation is expressed by # (from which, combined with &pa, interrogative dpa is probably derived): though, strictly speaking, the interrogation is not expressed by the particle, but understood: Ioveis rodro; ‘Do you do this ?' ’H motets root; «Do you really do this?’ (3) Besides expressing modes of thought, these particles, with some now to be mentioned for the first time, also indicate moods of emotion, nuances! Thus pathos is often suggested by 6y, irony or sarcasm by 84 and &94ev (sometimes by ye), interest and 1 tt may be objected that the particle merely emphasizes, while the emotional nuance lies in the context. But the particle, from constant use in a particular kind of context, acquires a specific emotional tone, @ INTRODUCTION xxxix surprise by dpa and ye, sympathy, encouragement, threatening hostility, and other attitudes by zor, sudden perception or appre- hension by kat pyjy and Kat 87. (4) These particles of emphasis and nuance I will style ‘adverbial’! since they are in most cases naturally translated by adverbs, ‘really’, profecio, certe, etc.: I shall apply this term to all uses other than connective and preparatory (apodotic uses are difficult to classify: see 5.d below). The contribution which these particles make to the force and vividness of Greek has been universally recognized. Often they cannot be appropriately translated into a modern language,? and their effect must be suggested by inflexiuns of the voice in speaking, or by italics, exclamation marks, or inverted commas in writing. It would be too much to claim that the whole expression that a sensitive and intelligent reader can put into a page of English is present already in the corresponding Greek, owing to the presence of particles. Rather, the particles may be compared to the marks of expression in a musical score, which suggest interpretation rather than dictate it. To carry the analogy further, a page of Thucydides bears somewhat the same relation to a page of Plato as a page of Bach to a page of Beethoven. (5) Hitherto we have considered the function of particles as expressing a mode of thought or mood of feeling in isolation. We have now to discuss their function as establishing a relation- ship between separate ideas. Relationships may be established in different ways, (a) The second idea is linked to the first by a connectins particle? which may do no more than connect, bus may also give * This is not a very happy term, but it is a convenient and customary one. Dr. R. W. Chapman, in some of his notes on the Greek particles, sty:es th.- Btoup ‘self-contained ', ‘independent’. We must include among the * Pendent’ particles the Epic re of habitual action, wich, froin Ks objective nature, stands apart from the other members of the group * German is richer than English in pasticles, and overs more equiva German writers on the subject start at sume advantage in this respect, _* Naturally, the units connected are normally eiusdens ori, Bus hs. is by no means a hard aud fast rule. E.g. in ASupp, aks an adver (Oa participial clause. Cases in which a finite verb is linked by acsxl INTRODUCTI@® a logical turn (adversative, causal, or inferential) to the con- nexion. (8) As expression develops, subordination largely replaces co- ordination, the Aéfis xareorpappérn the Aégis eipouévn, and to that extent hypotactic conjunctions replace connectives, These conjunctions, éme‘, e/, and so forth, must themselves be regarded as particles.!_ The only reason that I do not discuss them, as some other writers on the particles have done, is that their im- portance is grammatical rather than stylistic. (c) The capacity of particles to establish a relationship between ideas is not limited to the sphere of connectives and hypotactic conjunctions. xaé and ovdé* in the adverbial senses ‘also’, ‘also... not’, or, with a sense of climax, ‘even’, ‘not even’, point a reference to a second idea either expressed in the context or supplied by the imagination. Since this use of particles denotes that ‘one term answers another, I term it ‘ responsive’. (In this class we must perhaps include ody in its Homeric use, ‘in accordance with what I have said’, and éé in xal... dé, if (or when) xai is the connective in this combination (p. 199, n.1).) In certain cases this responsive use has a structural importance: as when, for instance, xa/ in relative and final clauses marks the addition of the content of the subordinate to that of the main clause. (@) Apodotic uses. Even in hypotactic constructions para- tactic particles (dAAd (dAX' odv), abrdp, O€ (8° obv), kad (Kal wiv), ovv, Toivvy) are not infrequently found at the opening of an apodosis. This apodotic use is probably a legacy from the earlier, paratactic, stage of expression, retained, perhaps, from a Greek love of clearness and logic: it is significant in this connexion that apodotic 8€ and xaé are common in Homer (the former also in early prose, Herodotus). But it is no doubt alternatively possible to regard the apodotic use of at any rate some particles as a relic of an earlier, adverbial, use, For it is by no means certain to a qualifying word, phrase, or participial clause are due to a form of ellipse common in all languages. A¢yeis, xai dpBis ye (Adyeis). Cf. ¢ (Dp. 164, (3), pirros (p. 406, Th.iv.51: ef. PLAp.2g¢), re(p. 502(g)). See further p. 497, n.2. } ‘The line between parataxis and hypotaxis is a very thin one where e.g. éxai or ds introduces an independent sentence, and thus virtually = yéip. See Kithner, 1 ii 461, Anm.t. * That re can ever mean ‘also’ is highly doubiful. See pp. $35-6. * | borrow this term from Hartung, but give it a narrower denotation. om INTRODUCTION xi that the’connective sense of any Greck particle is the original sense. Further, I class as apodotic the use of an emphatic particle at the opening of an apodosis: ye (yodv, with its negative form odxouy ... yevij (H mou, re), uévros (p. 402). Apodotic 84 clearly belongs to this category, for it makes its appearance long before 6% has begun to acquire connective force. Thus I include as apodotic both connectives transferred from parataxis to hypotaxis and purely emphatic particles like ye and 84 which do no more than underline the opening of the main clause. There is some illogicality, but great practical convenience, in embracing both types ina single term. Sometimes, indeed, it is difficult to distinguish one type from the other. Thus apodotic xal dy may be either connective or adverbial in origin (the latter, I think):? apodotic ody and adrdp, and apodotic re (if we are to recognize it at all) similarly admit of either explanation? In both its varieties, the apodotic use possesses a structural function in the architecture of the sentence, serving to stress or clarify the relation between clauses. For this reason I do not, for example, regard o¥8é in S.OC 5go (see p. 195) as apodotic. Here oidé is not, I think, ‘ the negative counterpart of 8 in apodosis ’ (Jebb). Rather, ovd¢ of seems to be the negative of xai cof (‘you also’, ‘you, on your side’) and the particle goes closely with the word that immediately follows it. The same consideration leads me to deny the title ‘apodotic’ to xaé cohering with a single word at the opening of an apodosis (sec p. 309). But the distinction is delicate, perhaps precarious. Closely allied to the apodotic use is the resumptive, in which 6€, 84, and ody pick up the thread of a thought which is begin- ning to wander. {e) The corresponsive use of particles, Coherence of thought is adequately secured by the presence of a backward-pointing Particle. The reader or listener, when he has reached a certain 1 An apodotic use of xai 8) cai, based on the connective use of that com- bination, appears occasionally to present itself, but crumbles to nothingness at the touch (p. 257). 2 For example, Kilner ([1 ii 327) regards apodotic ofy as adverbial, and the Homeric use of the particle gives him some support, An adverbial force is also clearly present in uirdp, if abrip = adre-+ dp.xiii INTRODUCTIOWy point, meets a particle which looks back to the road he has traversed, and beckons him on in a certain direction. But greater coherence is attained if in addition a forward-pointing particle warns him in advance what path he will soon have to travel, the connexion being expressed reciprocally, from rear to van and from van to rear. It is characteristic of the Greek love of orderliness and lucidity that this double method of connexion is already present in Homer. The forward-pointing particles, which we may describe as ‘ preparatory ’\! are wév, re, and xaé. Hév is most frequently answered by 6é, but often, too, by other particles: re by re or xaé, xa by xa‘, The mutual relationship between the earlier and the later particle may be expressed by the term ‘corresponsive’. In particular, the commonness of Hév ... 6€ in all periods of classical Greek has often been noticed. The tendency to view one idea in the light of another idea more or less sharply contrasted with it was indeed innate in the Greek mentality (and occasionally led to the employment of merely formal antithesis for its own sake).2. The result is a great gain in clearness and precision. Often, when a writer embarks upon a disquisition which appears to invalidate his own point of view, #& indicates that the aberration is only temporary, and that he will return after a time to the straight path? The responsive use of xaé and ovdé noticed above leads also to a corresponsive use in hypotactic constructions when the particle is present in both the subordinate and the main clause. Thus, Gonep cat éxeivor gira, obrw cat v2 PiXd: ‘as I love him as well as you, so I love you as well as him.’ The reciprocal relation between xaé and xaé is as clearly marked here as in paratactic construction, PiAd nai éxeivov xai oé. * A forward-pointing particle demands a backward-pointing one to answer it. Anticipatory yép is not strictly preparatory: it arises, as the word ‘anticipatory ’ suggests, from a dislocation of the natural order. Nor, con- sequently, can we class anticipatory ydp picked up by obv, 8%, etc, as corresponsive. * Demetrius, De Elocutione 24, in discussing this matter, quotes Epi- charinus' parody : rea piv év riveis yoy fv, rica 8é mapa rips éyar. Pear~ son, Fragments of Sophoctes, vol. ii, p. 298, observes: ‘The Greeks saw a contrast everywhere, and sometimes overdid it’, * Writers of Greek prose versiuns sometimes fall short of the Greek standard of lucidity in this respect. | 7 TR 4 e INTRODUCTION xiii Il. CONNECTING PARTICLES, (1) The origin of connectives. 1 have remarked above (1) that certain particles (d\Ad, mov, rot, and 7é in rovydp) can be traced back to other parts of speech. In other cases, where the derivation of the particle itself is unknown, we can trace the evolution of a connective from an adverbial sense in extant Greek literature. Thus wv and 84, in Homer affirmative particles, later acquire respectively (among other uses) an adversative and an inferential force." An adversative force of pév, hardly to be found in Homer, is later present in pévros, wiv obv, uly df. An inferential force of dpa and ovv, no more than nascent in Homer, grows to maturity. On the other hand ardp, yp, 8, eal, re can- not be traced back to an adverbial stage. But it is on general grounds probable (since the connexion of ideas, even in the simplest form, is mot a primitive process) that here also an adverbial sense lies behind. And the plausible derivation of ydp from ye &p (however little it may help us to understand ydp) points, if correct, in the same direction. (2) Connexion and asyndeton. As a general rule, Greek sentences, clauses, phrases, and single words are linked by a con~ necting particle ® to what precedes. Connexion is, on the whole, not often omitted in verse, stitl less often in prose. There are, however, certain well-marked exceptions to this principle, and Greek frequently dispenses with connexion in the following cases. (I will call this ‘ formal’, as distinct from ‘ stylistic’, asyndeton.) (i) The preceding context makes the connexion obvious, and to particle is required to point it. This is the case where a writer or speaker directly or indirectly announces his theme in advance, and where a forward-pointing pronoun or demonstrative adverb, or some other word or phrase (especially such an expression ) For tendencies which may have led to the evolution of connective 34, see Bj, 1V.1, ad init. * The line between connectives and non-connectives cannot de rigilly drawn, Thus oy in Homer, although it has not yet developed a connective fanction, shows in piv ody a tendency to develop one. y¢, and in a more marked degree wiv ye, mitigate to some extent the harshness of an asyndeton: while yooy in the * part-proof” usage is almost a full connective, per, again, occasionally appears to have 2 quasi-connective force {p. 360).xliv ee as rexpypioy dé, onuciov 8¢), supplies the limk. E.g.Th.vigo.2 ... yddere fbn, éwdeGoaper . . 1 Dwvit7 doyiferde ydp. dpyew Bovrcrar: xv... ray yeyernpiven Ouds Th... Uropriiow, dueTs eLeméuypare ...: Thiili 20.3 évéueway ri €f680 LOedovrai rpbry rose. KAipaxas éroijoavro. But in such cases connexion by explanatory ydp is probably commoner than asyndeton. (ii) To a less degree, a backward-pointing pronoun or demon- strative adverb, usually at or near the opening of the sentence, similarly diminishes the necessity for a connecting particle (8%, obv, or roivuv), E.g. in X.dm.iz over twenty sentences begin with évrai0a, évrei6er, ravrny, etc., without a connecting particle. Occasionally the pronoun is placed comparatively late: Thaiif 28.2 pay 2ipBaais abry éyévero: Ant.vit4 xabeorixer piv 4 xopnyia obrw: X.dni89: Anditg: D.xviiia35: xx 55. (iii) In a long series of co-ordinated nouns, adjectives, or verbs connectives are, on the whole, more often omitted than inserted. Th.iig.2 Meyapijs, Bowrot, Aoxpol, Boxis, 'Apmpaxiarar, Aevad~ Sot, Avaxrépio: Pl. Phdr.2538 (adjectives), But sometimes, and in Epic normally, connectives are inserted: the ancient critics styled this polysyndeton. Thus, Hes.74.205-6 and 320 (series of re): #6.243-62 (re and xaé alternating): Th.iiitor.a Invéas xal Mesoanious cai Tpiraséas xai Xadaiovs xai Todropeviovs cab ‘Hogious xai OlavOéas: Hdt.ivi02: PI.R.618D: Lg.758E,842D, 9428. (Our convention of linking the last two units only, leaving the rest unconnected, is on the whole alien to Greek usage: see O€ (p. 164), kal... 8€ (p. 202), cad (pp. 289-90), re (p. 501).) PI.Zg.8974 is a good example of varied asyndeton and connexion. In a negatived series, while the employment of asyndeton with- ont repetition of the negative is not excluded, repeated negatives, with or without connectives (oUre.. . ore... ob re: ob... 0888 «+» o86é: ob ...00... ob) give an effect of greater force, by elimi- nating each item individually (‘not A, nor B, nor C’: ‘not A, not B, not C’), instead of eliminating the entire series en d/oc (‘not A,B,C’). E.g. Pl.Lg.832¢ (5 odre's) : 8y8B (8 #ndé's) : 902D,935B. In Pl.Smp.2114 the great series of odre's and oldé's, enumerating one thing after another that true beauty is xof, and leading up to the revelation of what true beauty is, dAX' adré xa’ abrd ped’ atrot povoesdis dei bv, lends an astonishing power and passion to the period. Cf. Hom. 36y~g2, where the oddé’s (connective and eer A BEES et ere pm etme INTRODUCTION xtv adverbi 9... like hammer strokes: ‘No, no, no!’ In English, compare St. Paul, Romans viii 38-9 : ‘For I am persuaded, that neither death, nor life, nor angels, nor principalities, nor powers, nor things present, nor things to come, nor height, nor depth, nor any other creature, shall be able to separate us from the love of God, which is in Christ Jesus our Lord.” The truth of the matter is that a great chain or series is of its essence impressive, whether connectives are inserted or omitted. Whether asyndeton or polysyndeton is the more impressive in a particular place, depends on the nature of the context. The Greek critics rightly regarded both as rhetorical ‘ figures’. Stylistic, as distinct from formal, asyndeton is used, sparingly by some writers, freely by others, for emotional effect : the im- pression given is that the speaker's or writer's feelings are too deeply engaged to allow him to trouble about logical coherence. Longinus has some admirable chapters (19-21)on this subject. He quotes Hom.x251-2 as an example of asyndeton in rapid narra- tive, and acutely analyses the telling use of this device in D.xxi72. In verse there is no finer example of the effect of asyndeton, combined here, as often, with repetition, than the lines of Aeschylus quoted by Plato, X.383B. Of prose writers, Thucydides and (of course) Isocrates rarely employ asyndeton, while Demos- thenes exploits its possibilities to the full. As Aristotle remarks (RA.1413b17-31), asyndeton is essentially a dramatic device, and is for this reason appropriate to oratory: one must ‘act the passage, not merely speak it’, A good instance of accumulated asyndeta is to be found in D.xxiv 11-14, where, to say nothing of clauses, ten consecutive sentences, covering twenty-seven lines, open without a connecting particle. Out of many fine Demos- thenic examples I will cite xviii67,299 (the latter perhaps the finest of all), and xix 76 (where asyndeton expresses the stunning rapidity of disaster). In Lycurg.33 asyndeton gives pithiness : ri yap ter mpopdceay } Nbyov } oxryews; dwAodv Td dixacov, Aadtov 7b dAnGés, Bpaxds 6 Eeyxos. In contrast with this stylistic employment of asyndeton Andocides and Xenophon often omit connectives in narrative with a certain naive awkwardness, and without any apparent thetorical justification. E.g.And,i41,42,82,120,1a3: Xtal 2.33: iv 5.33: v6.25: viga8: Cyriitr8 :xivi INTRODUCTI (3) The mode of connexion omitted when stylistic asyndeton is used is in most cases ‘and’. Less frequently ydép or yoiv has to be supplied, as in E.Or.234: D.xviii2g9. Sometimes ‘ then’ or ‘therefore’ has to be supplied : PL.Pr#.339E ods pévrot Sipw- vi8ys wodirns’ Sixatos ef BonBeiv 7G dvbpl: Pi.0.3.45. (4) In deciding whether asyndeton is tolerable in a particular place, the usage of the author and the charactcr of the passage must be taken into account. These considerations are sometimes of importance for determining the text. Thus in X. Hier.6.6 (p.551) 6 yé roi PSBos «rd. (re 4CM) a connective seems needed to mark the introduction of a new point: the yobr sense of yé ros will not suit, and the analogy of An.vis.24, where ydp has to be supplied in thought, does not support the asyndeton here. d€ rot (Bach) should perhaps be read. Again, in P1.Grg.459a (p. 578) Eneyés roe vuvdy xpd. a connective is badly needed ; nowhere else in this passage of formal dialogue (458E-459C) are the success- ive stages in Socrates’ argument introduced without a connéctive (ye in 4598 carries on the thread from his previous speech). I believe roivuy wv 8H (roivyy viv d1 ) to be the right reading (C.R.xlvii (1933) 216). In E./7'50 Porson’s emendation pro- duces a most improbable asyndeton: 14.1175 the asyndeton is difficult, and Paley’s defence of it hardly convincing: in HF 722 I believe Nauck's (8°) to be necessary. The question of the permissibility of asyndeton is of vital im- portance for the true explanation of anticipatory ydép. The asyndeta resulting from the view that ydp is adverbial here are often intolerable : see ydp, IV, and dAA& ydp (p. 100). (5) Apparently superfluous connectives. In certain cases con- nectives are inserted where they are, strictly speaking, unneces- sary. (i) In answers to questions. wév odv (with preparatory pév and connective obv : ‘ Well’): PL.PAlb.518: Sph.22gd: X.HG vi3.13 (answering a rhetorical question). 8: Pl.Chrm.172c dpa.. Tdya 8 dv, én, obras éxo (* And perhaps it may be so’): often in answering a second question (pp. £71~2). In answers to commands, Pl.Euhphr.i5& ent.» ..—Eis aibis rotvuv, & Zéxpares (* Well, another time, then’): 2.3370 dnbrecov dpyépiov.—Ovxotv tmeddv por yevtat (‘When I get INTRODUCTION xvii some, nc, Fora curious rofvuy in a rejoinder to a state- ment, cf. Ar.V.t141 (p. 573). (ii) At the opening of a speech or oracle, or of a whole work. The explanation of this inceptive use of connectives is perhaps not everywhere the same. Often the speaker wishes to put his thoughts into relation to the view of the persons he is addressing, or what he takes to be the generally prevailing view. But often again, this use of connectives appears to be a mere mannetism of style. It has always a touch of naiveté such as is characteristic of Xenophon. See dddd (11.8), 8 (I.C.2.ii), rotvur (1.6). pér is similarly employed in openings: though not a connective, it seems to mitigate the abruptness of the initial plunge (p. 382). In a political or forensic speech, after the recitation of documents, the practice varies, connexion being usually inserted, but some- times omitted. {iii) In reported speech an opening connective is naturally omitted. ‘He said,“ Then I'll come”’ becomes ‘ He said he'd go’. But there are cases where the connective is retained: kai, X.HGv3.10: 3.15: yyy, E.Tr.1138: ody, PLPrtgarc (p. 426): roryapav, Hdt.iv 149.1 (p. 567, n.x): rotvuy, X.Cyrvi 3.17 (p. $71) For possibly superfluous 8¢ in exclamations, see p. 172. (6) The different methods of connexion. These are, broadly speaking, four: (@) Additional, (4) Adversative, (c) Confirmatory, (2) Inferential. But the di ns are everywhere fluid. (a) represented at its purest by xa/and re,(though 6 is often hardly tinged with adversative colour): one idea is simply added to another without any indication of a logical relation between the two! A variant of (a) is what I shall term the ‘ progressive '? use of particles, or combinations of particles, conveying not merely the static piling-up of ideas, but movement of thought : ‘now’, ‘again’, ‘further’, ‘to proceed’: eg. pry, GANG wre, ye * The logical relation may be inherent in the context, though not expressed by the particle, xai, like ‘and’, sometimes stands where ‘and yet’ is implied (pp. 292-3). So, occasionally, re (p. 514 én), Again, i, mean- ing ‘and’, is sometimes used where the logical retation woud property be expressed by yin, ody, or F (pp. 169-71). * E prefer this term to ‘continuative ', which some writers have employed, A man going round in circles in the desert ‘continues’, but does not eo gress’, on axlviii INTRODUCAGON piv, xal piv, pévro! The same significance may be reached from the direction of (d), when ody and ovxodv degenerate from propter hoc into post hoc. In this progressive sense particles mark ‘something of a new departure in the march of thought. They convey an effect approximating to that produced by paragraph- ing, though not usually denoting quite so strong a break. An example ex contrario will illustrate this, In PI.R.338A Edrévros 8é pov taira starts a new paragraph in the Oxford Text, and is printed, as here, with an initial capital. At such an important joint in the structure ody or rocyuy would have been more normal. (Cf. the not infrequent use of 6 in resuming after paprupiat : eg. D.xlvii52.) It goes without saying that particles, when used in the progressive sense, must follow strong stops. But certain particles and combinations regularly so used occasionally follow weak stops: e.g. 34 (p. 239), dAA& 87 (p. 242), xal priv (p. 352), roivv (p. 577(5)). kat 6%, wal... 6, and al 6} kai occupy a position between the purely additional and the progressive particles. Broadly speaking, they are to be classed with the former. But xai 6 sometimes introduces a new point, like xat Hiv (p. 249), and so, rather more often, does xai 6 xai, which tends to follow a heavier stop than xal 64 and cal... 3. Kai... 8é, on the other hand, is rare after strong stops (p. 201). Even among the particles which I have described as denoting addition pure and simple some difference of structural function can be detected. There is a certain tendency, I think, to use dé, rather than xaé, for connecting sentences {in the same way as many English writers avoid ‘ and’ at the opening of a sentence), while it is hardly used at all for connecting single words (p. 162). The case of re is complicated, some writers preferring to use it for joining sentences, others for joining clauses, phrases, or single words. The line between additional-progressive and adversative is 4 Even within the limits of this class a certain distinction may be drawn, in the uses of such combinations as xui piv and roivyy, between the mere transition to a new item in an enumerative series, or to a fresh argument, and the arrival ata new stage in the logical process. The former may usually best be rendered ‘again’, ‘further’, the latter, ‘now’, ‘well’. 9 The post hoc sense is clearly the later in ay, and also, J think, in obxodv, In rainy, on the other hand, if Wackernagel's etymology is right (p. 568), the proper hoc sense is the later. Aeon ome oe ee a INTRODUCTION xlix more sharply drawn in English than in Greek and in Latin, aanrd, drdp, adrdp, 8, wry, GANA phy, ye why, eal phy, xairor pévros, etc., like a¢ and axéem, are used both to add and to con. trast.|_ On the one hand, the adversative force of a particle like dad is at times weakened : on the other, custom attaches an adversative force to a pure connective like airot,® or to an originally emphatic particle like piv, or, in certain combinations (ute 44, wey obv, pévror), to pév. By these new developments the range and variety of adversative expression is considerably increased in post-Homeric Greek. (6) Adversatives are of two kinds : eliminative ? adversatives, used often where one of two contrasted members is negative, the true being substituted for the false (par excellence pee oly and normally ddd), and balancing * adversatives, where two truths of divergent tendency are presented (8é, wjv, pévror, etc.). Inter- mediate between the two we have adversatives like drdp and xairot (and sometimes dAAd) which simply raise an objection, leaving it uncertain whether the objection is a fatal one or not, These dis- tinctions are important in principle, though the dividing lines are everywhere fluid. For example, the readiness of d\Ad and 6 to exchange functions is illustrated by the abnormal, but not uncom: mon, use of dAd to answer yév, and of 8 to contrast a positive with a negative clause. yjv, kal priv, and pévror occasionally approach the eliminative force of pev obv (pp. 335. 358, 403). Class (c) is represented throughout Greek literature by ydp alone (though yodv often approaches yép in force, giving partial confirmation). (d) is less prominent in Homeric Greek, since "When dana iv, xal jjr, xatroy, 8 ye, etc., introduce the second (major or minor) premise in a syllogism, it is often difficult to say whether they are adversative or progressive. See xuirot, p. $63, and Addenda to p. 353. * Adversative kairo: is so common that one may legitimately reckon an adversative sense as one of the senses of the particle. On the other hand. where xui appears to be adversative, the opposition is inherent in the con- text rather than expressed by the particle (p. xlvii, n.1), Decision between these two explanations is often a delicate matter, We should not, { think, Tesort to the second in the case of well-established usages. Thus | cannot agree with des Places when he says (p. 107) that in corrective piv ole Vopposition reside uniquement dans la pensde, non dans la particute * T think that these terms express the essence of the distinction better than ‘ strong" and ‘ weak *,1 INTRODUCTEQN supplying an explanation is a more primitive and natural process than drawing an inference. In Homer (d) is represented by 73 and rotydp, the inferential force of odv being still in embryo. Subsequently 7 almost entirely disappears, while rovrydp remains (in prose almost only in the strengthened forms rovyépro: and rotyapodv), and additional inferential particles are found in ob, 6%, dra, dpa. (7) Abnormalities of reference in connexion, The connexion established is normally, of course, between consecutive units of speech: words, phrases, clauses, or sentences. There are, how- ever, certain exceptions. In dialogue, owing to the quickness of thrust and parry, or the self-absorption of one of the participants, a speaker sometimes links the opening of his speech to his own preceding words, not to the intervening words of the other person.! Thus S.07 1357 (ofkouw ... ye): P4.1257 (kaérot): E.Hel.1259 (ye piv 64): Or.793 (bv): Ph.608 (ye) : 4€ ye (p. 154): perhaps €f ydp in A.Pr.1527,C#.345 (p. 92). This is often the case with ydép (IIL.5). In S.E7.1035 (p. 443) 4AA’ obv looks back to 1017-26: or perhaps it would be truer to say that its point ad’appui is the general situation, the whole attitude of Chrysothemis, rather than any particular set of words, an explanation which applies also to E.AL.713 (cat piv, p. 354), and 77637 (pévrot, p. 405).2 yap (III) presents, in general, many abnormalities of connexion. Thus it sometimes refers to the motive of the preceding words (not to their content), to a far-back remark in a continuous dis- course, to an individual word or phrase, or to an idea suggested rather than expressed. Sometimes, again, two successive ydp’s share a common reference. ! The ignoring of this possibility has sometimes led to misunderstandings. A certain flexibility of mind is required in such cases. Jebb on S.£/.1035, Paley on E.A/c.713, have, | think, interpreted the sequence of thought too rigidly. ? But see Addenda to p. 92. + In some of these examples the opening of the second speech is marked by a particle which 1s not, strictly speaking, a connective. But the line between connectives and non-connectives is a shadowy one (see p. xliii, n. 2), and the principle illustrated is the same in both cases. ® INTRODUCTION hi Ill. COMBINATIONS AND COLLOCATIONS OF PARTICLES. (1) The distinction between combinations and collocations. There has often been occasion in the preceding pages to cite com- binations of particles, It is now time to consider how far particles may be said to cohere so as to form a real unity of expression, as opposed to a merely fortuitous collocation. (i) The combinations which have the most indisputable claim to the title are those in which one or other of the two particles (either the more or the less important of the two) could not have been used without the other, and also bears in the combination a sense which it cannot bear in isolation: for example, adverbial ody reinforcing dAXd, yp, ye, 8é,or preparatory pév : adversative pe preceding 84, obv, or ror. In corrective pév ody neither particle could have been used without the other, and each bears in the combination a sense quite different from its independent sense. (ii) In other cases, while neither of the two particles bears an unwonted force in the combination, still, the presence of the one depends, at any rate to some extent, on the presence of the other. E.g. Pl.Lg.666A per& 8% robro (vopoberjiaoper) oivou piv 8} yedeoOar Tod perplou ... wéOns d%... dwéyer Bar. Here it might at first sight appear that 4 does not adhere to #4, but emphasizes ofvou, or the phrase ofvov ... yeveoOas. But. as we find that in prose 6% rarely emphasizes substantives or phrases, while it very frequently follows preparatory per, there seems no doubt that the function of 4% here is to stress the anti- thetical form of the sentence, not to underline a single element in the content of it. Similarly, yap 6) usually? denotes that what is presented as a cause is in truth a fact: ‘for actual The case of ye pv is instructive, From the point of view of 1 Thave not seen this question discussed in any treatment of the particles. My thanks are due to Dr. Chapman for urging me to clear up my Weas on the subject. I have not, however, attempted in this book to use the distinc tion between ‘ combination’ and ‘collocation’ as a basis of classitication, To have done so would have complicated matters needlessly. * For exceptions, see LV below.lit INTRODUC(JON abstract analysis, ye emphasizes the word it follows, and pv is the connective, both particles exercising their forces independently, But ye piv is used in positive adversative clauses by writers who use simple adversative yyy in negative clauses only. The association of ye with yyy is therefore stylistically important, and the two particles may justly be regarded as forming a real com- bination. So, too, may | ydp and jj xa‘, which are used by Plato and Xenophon far more freely than # simplex. On the same prin- ciple, in Attic, where connective xai 6% without a following xaf is rare, eal 6} xai must be regarded as a combination, whereas in ai piv Kai, where the addition of a second xaf is not pre- scribed by custom, there is less coherence between the first two. particles and the third. (iii) Sometimes, again, while either particle could stand with- out the support of the other, the two nevertheless tend to cohere. Thus, with xa? ydp, though ead often goes closely with a word following ydp, there are cases where xai seems to bear upon the sentence as a whole, and to cling to yép: ‘for there is a further fact’, Again, the very frequent occurrence of xai 6} xai in Herodotus seems to suggest that, although he, unlike Attic writers, freely uses connective kai 5 without a second xai, still even in him the second «ai, where it does appear, is an integral part of the combination. (iv) In other cases the collocation of two particles is purely for- tuitous. Forexample, I see nothing significant in the frequent juxta- position of preparatory yév with ydp and roivuy:? and if I mention in my text those uses of wiv 64, uév obv in which the first particle is preparatory and the second connective, I only do so because of their bearing on the evolution of connective 6% and ove, and in order to call attention to the importance of distinguishing between two entirely separate usages. This is perhaps the place to mention the tendency of certain particles to gravitate towards certain other words which are not particles, especially towards pronouns. Thus ye, especially in Homer, but also to some ex- tent in later Greek, tends to attach itself to pronouns (pp. 121-2) : * See IV below. * Des Places (p. 308) attributes stylistic importance to piv roisw, and Shorey (C.PAi/axviii2) calls attention to the frequency of piv yép in PLP rt. 337A. & — @ INTRODUCTION titi 0 does emphatic pév in Homer (p. 360). pévror in its affirma- tive and syllogistic senses is frequently associated with 20, odros, rotodros, Tovdade, etc. (pp. 400, 408-9), ye tends to follow (often with a word or words intervening) ds, ei, Srav,? ete, (p. 151,n. 1). 8% tends to adhere to certain words and types of word: eg. to adjectives expressing indefinite quantity or number, to viv, to bios, to superlatives, to dpav (pp. 204-18). yay is predominantly used after negatives (pp. 330, 334-40). (a) Avoided collocations.* Certain collocations of particles, which are in themselves natural enough, are for some reason or other wholly or generally avoided. rofré ye yap odx épeis seems as unob- jectionableas ror6 ye phy ovx épels. But while ye pv is not merely a tolerated collocation but an established combination, ye ydp never, I think, occurs (the analogy of ye piv tells strongly against the explanation I have offered on p. 151). The derivation of yép from ye dp might perhaps be the cause of this avoidance (just as 8} dira iseschewed). For though yap dpa is found, a language may tolerate a harshness in one case while avoiding it in another. But I doubt if the Greeks apprehended, whether consciously or unconsciously, this derivation of yép (if, indeed, it is the true one). The similar avoidance of ydp ye ® is more easily understood, as it violates the normal order of precedence (see V.2.i below). While ydp mov is common (in Plato), 8 wou and xai mov not rare, and aAAa pijv ov, cai priv mov occur several times, wou never seems to follow an inferential particle such as connective 64, obv, Toivuy. re obv (re prospective and ovy ancillary), in marked contrast with eize obv, odre obv, is only found once (p. 420), while the colloca- tion of prospective re with connective ody is almost entirely avoided, except by Plato (p. 441). wey rot (separatin, with pre- paratory yév), 7é 701,4 and ody rox seem to be avoided, in contrast with yé roi, 8€ rot, yp Tov: so, on the whole, is TE ye (P. 161, * 1 do not include these conjunctions among particles. See 1.5.3. Nor do T include the negatives, of and pj. * See further LV below. , * In E./on 847 yap ye, which Gregoire (in the Bude edition) surprisingly Fetains, is generally held to be corrupt. * Conjectured by Buttmann in S.P4.823.liv INTRODUCT@N and see IV below). re wév (re = ‘ both"), which might have been expected to occur sporadically,! seems not to be found. Other avoided juxtapositions are od (ur) and ye (p. 148), ef and te simplex, ov and preparatory pév (in that order)? See also (3) below, ad fin, In some cases it is the toleration of an apparently harsh or awkward collocation that calls for remark. Thus non-connective kal 6% and corrective péy of are juxtaposed with each other, and with other particles (pp. 250-3, 479). (3) Split combinations, Particles may form a combination even when not juxtaposed.? There is no distinction in meaning between é¢ yeand 82... ye,dANd ydpand dAAd...ydp, kai pévrot andrat...pévrot. In certain passages yév and ye separate but in close proximity have almost the same effect as wévye. pev...toivuy is occasionally used for pév roivuv. In poetry metrical considera- tions often tilt the balance towards juxtaposition or separation, In other cases custom changes, or individual preference plays a part. Thus Homer writes xai... mep, odd? . . . ep, later authors xainep, ov8é mep: GdA& ydp replaces dd\AG... ydp during the fourth century : Plato prefers cai wévrot, Xenophon xai.. . pév- rot. Sometimes the juxtaposition of logically cohering particles is actually avoided : notably in the case of ye following an earlier particle or combination (p. 152). * Plato writes (Zg.655A: See p. 373) aAN' ey yap povorey) xai oxipara per rai pidg for... If he had used re... xai instead of rai...» xa, the passage would have run GAN’ év yap povosed aximard re pév Kai wédn treats... That nothing of the kind turns up in the whole of Greek literature is perhaps not accidental. re piv, though logical enough, would have been felt to clash. Even as it stands, the sentence is exceptional. The intrusion of per between the corresponsive cai's is awkward, and the natural order would have been 1 pév aires raw avoids dnare wiv oxipurd re ani pidg. In A.Ag.3y6 deriv 8 Dardv 86 piv buoies. re pév (re = ‘and"') is found in Hp.Mord.ii53 (p. 373). But it looks highly suspicious. 7 Conjectured by Wilamowntz in $.P4.811 (p. 331). This taboo is evi- dence against taking wiv in od wiv 0j in PILPAlb.468 (pp. 292-3) as pre- paratory. My Ocford votleagues tell me that their pupils frequently write ob war. * But this does not apply to ai! combinations. Thus dA\' odv is never split, and Jebb is certainly wrong in associating the particles in S.dntg2s (see P-473). cai piv is hardly ever split (p. 354). @ INTRODUCTION wv I have referred above ((1) ad fiz.) to the association of ye with conjunctions, @s, ei, 8ray, etc. Here, again, the association is not dissolved by spatial separation, and it appears to make no difference to the sense whether ye follows at once or after an interval (pp. 142-3). The juxtaposition of ef and ye seems to be mainly, that of ds and ye (with ds = ‘for’: p. 143) wholly avoided. (4) Exceptional combinations. Generally speaking, a combina- tion cannot rightly be described as such unless it is more or less established in the language. Occasionally, however, we meet with an ad hoc combination, in which, though it may be found only once in Greek, there is a close and essential cohesion between the separate parts: thus, od piv dANa... ydp (p. 30), GAG phy... ydp (p. 347). (5) Double connexions. In a combination of particles it is normally the case either that one particle (whether the first or the second) is connective or preparatory, the other adverbial (e.g. re O%, ye wiv, yap ody), or that both are adverbial (e.g. ye 4%, (yap) obv 6, } To: pév). For the order of precedence, see V.2 below. In a few cases, however, each particle appears to be connective, the connexions being of different kinds. See d\A& yép, (9) and (10) (pp. 107-8): yap ... dra, kal... Sra (pp. 272-3): GAG... rolvuv, S&... rotvvy (p. 579). Adversative kai piv, kat pévrot, should not, I think, be so explained, tnough the analogy of ‘and yet’ is tempting (pp. 357, 415)- * Perhaps this is putting it too strongly. Certainly Sophocies aways separates «i from ye. But there are many examples in other authors of « y« tiv ye, iy yer ey F719: Orttob,tga3: Ar.Ardog: Anuph. Fr. Hdtiii73: PlLuigec: Pidr.2g2u: 253¢ (conjectured): N.Mesii Demosthenes, see Preuss’s Index. (In 1.3.228 Biass conjectured « y° fo de.) Wilamowitz is therefore wrong. in saying (on: f fom847) Sos dye tt une tulissig, da man in alter Zeit «i und ye durch ein Wort trent’ On the juxtaposition of ye and dy, see Neil, Appendix to Auayacs, p. Pearson on E.PA.1215,Wi INTRODUG (ON IV. DIVERSITY IN THE USAGES AND MEANINGS OF PARTICLES. We have seen (and the pages of this book will prove it abundantly) that few Greek particles possess one meaning and one alone. New uses develop out of old, and the old, though they sometimes wither and die, more frequently prolong their existence, often in altered forms, by the side of the new. The meanings of particles, more than those of any other part of speech, are fluid. mdvra pei. Some, indeed, like dA\d (and, with all its detailed subtleties, xa¢) remain more or less true to type through- out their course. Others behave more eccentrically, and of these the most unaccountable is re, whose two main currents no philo- logist has traced convincingly to a common source. Even in usages which appear rigid and stereotyped, the possibility of unexpected deviations from the normal has to be borne in mind.’ Thus, while e¢ xaé usually means ‘even if’, there are places where ‘even if’ makes nonsense, and xa‘, detach: ing itself from ¢/, adheres closely toa following expression (p. 304). xai following interrogatives is of three distinct types (p. 312). Tovrov kat Aéyw can mean either ‘He is just the man I mean’, or ‘I do mean him' (p. 322). ef ye is usually ‘if, but not other- wise ', but sometimes ‘even if? (p. 126). jjv has perhaps different meanings in the apparently similar idioms 7/ wv ; and dAXa ri piv; (p. 333). ob8€ in Herodotean of yey oddé has not always the same force (p. 363). In combinations of particles the possibilities of ambiguity are naturally increased : all the more so since, as we have secn, spatially separated particles may logically go together, and, as we shall shortly see, juxtaposed particles need not necessarily go together. In xal ydp, oS yap, xai ydp To there is nothing but the context to show whether (as usually) ydp is connective and xaf or ob8é adverbial, or vice versa (pp. 109-11). Similarly, in progressive aAAd ydp (p. 105), yép seems to be adverbial, which in this com- bination it normally is not. While in yap 4% the coherence of the two particles is usually beyond doubt (Hdt.i34 fv yap 8) cass, » Some scholars have gone astray in discussing $.07219-21 through assuming that od yip dy, which offen means ' for else’, ust mean that. See Jebb. Qa INTRODUCTION Iii ‘For in point of fact he was deaf’), in D.xxi4y they clearly do not cohere: for ri yap 84 mor’... .; is followed by xat méduv rf 84 mor’. ..; and in both cases 64 must go with the interrogative, and strengthen it. Similarly, in xa’ ydo, xaf sometimes refers to a following word or phrase, while contrariwise in xat .. . yép the particles sometimes, though separated, cohere, Where xaé, at the opening of a sentence or clause, is followed at a short interval by a second particle possessing both adverbial and connective functions, there are two possibilities, Either (as usually) xaé gives the connexion, and the second particle is adverbial: or xaf is adverbial, and the second particle is connec- tive, Thus in X.Mem.iv 7.4 xaé means ‘also’ and pévrou ‘ but’. raf is also occasionally adverbial in cal... 84 {p. 255), kal. roivey (p. 578), and, probably, kal... . 6€ (p. 199, m. 1). That two particles form an established combination does not mean that in no circumstances whatever can they part company and exercise their functions independently, (It is easy to go astray over this matter in réading. The eye catches the juxta- position, and the brain assumes a logical coherence. In some cases, if the passage were spoken, the ear might detect the dis- tinction by a stight change in inflexion.) In yody, for example, while odv usually stresses ye, there are cases where oly is con- nective, detached from ye, which goes closely with the preceding word,! In yé ror the stereotyping of the * part-proof’ sense (as in yodv) does not preclude the juxtaposition of the two particles in independent capacities (p. 551). In ob yap AA we sometimes find d\Ad separated in sense from ob ydp: ‘No, but...’ (p. 31). Here, obviously, pronunciation would indicate the grouping of the words, and a comma after ydp would make all clear to the eye, In one passage a similar division of o¢ pévroe dAAd is possible (p. 405). When we find a Greek author using a collocation of particles which the language in general avoids, we shall often find on closer examination that there is nq real coherence between the two particles, Thus, the only instances of 8 ye which appear to be sound are those in which 61 and ye do not coalesce (p. 247). * Perhaps in such cases y' ody should be written, separadiaz, to mark the distinction, See p. 443, & 1.tii INTRODUCTI@N The same is probably true of ré ye (p. 161). In Pl.Grg.454£ the ye goes with dAAa mrjv, in Phd.59C it is epexegetic of vat: ré ye is far less objectionable in these two passages, and in X.Mem.i2.54, than in E.d/.647 and Pl.PAd.1060, in both of which the ye can only be taken in close conjunction with the re, Similar considera- tions justify the rare juxtaposition pévro: ye in two Aristophanic passages (pp. 404, 410). Passages in which two particles normally forming a well- established combination exercise their forces independently are discussed on pp. 132 (o¥8é ye), 153 (8¢ ye), 159 (kad... ye), 160 (wer ye), 245 (ye 57), 402 (ye wévrdy), 412 (ye perros, perhaps), 433 (ye pévrot). Different meanings of the same particle or combination are even found in close proximity. Thus it seems likely that in Hes. 0p-.772 ye pév is adversative, while in 774 the pév looks forward to a dé (pp. 387-8). In Hdt.iary re 6% is apparently ‘and’ in one place, ‘both’ in another. In PI.Grg.§03B-C the first Ti 8€; is, I think, ‘Well’ ( And what of this ?'), going on to a new question : the second is a surprised question, with no connective force, + What?! In Enthd.2g8b the first Kal... ye is ‘Yes, also’, the second is‘ Yes, and...’ In $.0C539 and 546 Ti ydp; bears different meanings (pp. 82-3), in Ar.Vw.254,255 Tofvuy (‘then’ and ‘now’: p. 574), in Hom.M 344,357 wév (p. 368). In the last example we have different meanings not only in close proximity, but in identical phrases: cf. the different meanings of pév in 292, 8318! (p. 368), and of wep in Agi, T155! (p. 445, 0. 1). Cf also xai pay in Antv 44 (p. 358). Vv. THE POSITION OF PARTICLES. (1) The position of particles in sentence and clause, Adverbial particles, especially when they apply to the sentence as a whole, tend to gravitate to its opening,’ where the emphasis in Greek usually lies. 7 is in fact almost tied down to the position of "In these two cases the variation of meaning may be ascribed to the repetition of a stereutyped phrase in different contexts. ® Neil (A’mights, p. 186) has called attention to this in the case of ye + to ee a re INTRODUCTION lix first wora™in the sentence! The position of certain enclitic particles is further affected, as Wackernagel (/udog, Forsch, 1 (1891) 333-436) has shown, by the general tendency of enclitics to come second in the sentence: a tendency strongly marked in Homer, but considerably modified in later Greek. Thus in Hom.NIIP ov occurs fourteen times, always as the second word. Epic re and a also occupy the second place, and so does rot, which even in post-Homeric Greek presses to the fore so insistently that it sometimes cuts a compound verb in two: E, Or.1047 "Ex tol we ries? @qv in Homer, and usually in Theocritus, comes second word, except where two particles precede it (e.g. xal ydp Ov), But where the emphasis or tone of nuance is postponed, the particle is postponed with it, and ye and 6% sometimes come near the end, or actually at the very end, of asentence. Even particles like ror and mov, which bear on the general thought, are sometimes postponed. Thus in D.xviii 117 difrov ends a sentence : for postponement of roc and affirma- tive pévrot, see pp. 547-8, 400-1. Particles which affect the thought as a whole are comparatively rare in post-positive subordinate clauses. (Where the sentence opens with a subordinate clause, a particle contained in that clause is often to be regarded as belonging to the following main clause : ef roe radra Aéyers, perder = Wevder rou, ef radra déyers.) But we find yoow, for example, in a post-positive relative clause in Pl.Grg.goga : and rot (9.2, ITT) is quite common in post-positive subordinate clauses, Where ye (as usually) or 4% (as often) emphasizes an individual word, it normally follows it immediately, while xai (‘also’, + Interrogative j and dpa naturally open a sentence, or at least a clause: but there are exceptions. Wackernagel (/mdog, Forsch. i, 377) suggests that the non-enclitic ‘ post-positive' particles dpa, wip, , Sina, uev, wi, oly pers haps gradually became post-positive, like en#m, and sumgue on the analogy of enim, itague on the analogy of igitur, 3 in Homer can open a sentence, but is beginning to be post-positive. (roivw Wackernagel rightly regards as formed by two enclitics, rot and vw: see p. $68.) In rocyapoiy, I will add, We can watch a particle becoming post-positive (see VI. 3 below). The con. verse process is to be seen in dpa (in the sense of dpa), which writers of the New Comedy sometimes put at the opening of a sentence ip. 48, n. 3). "On particles in unesis, see Kubner, 11 i 530-7, and cf, pp. 420-38 47 (iv): Archipp.Fr.35.2).cere UN ‘even ’, ‘actually ') immediately preathes the emphasized word. But there are many exceptions in verse, and some in prose. Thus the most emphatic word sometimes does not immediately follow rai, and sometimes follows, instead of preceding, ye or 8%. The position of connectives is, naturally, far more definitely fixed. Kai, rorydp, rorydpro: always, rocyapoiv almost always,! occupy first place in clause or sentence. Other connectives norm- ally occupy the second place. The main exceptions to this rule are: (i) Postponement after closely cohering word-groups, parti- cularly where article, preposition, or negative (or more than one of these in combination) cling tenaciously to a following word. (ii) Postponement after an apostrophe, oath, or exclamation. (iii) In verse, postponement due apparently to metrical con- venience, Here the practice of different authors varies consider- ably. Thus Aeschylus postpones 8 more freely than Sophocles (though Sophocles often postpones of») and Euripides, and they more freely than the comic poets: while per contra the postpone- ment of ydp goes to surprising lengths in Middle and New Comedy. (2) Order of precedence in combinations, (i) An adverbial particle attached to a connective usually follows it, either im- mediately or at a short interval: yap 6, dd’ obv, kai 8, cal 6%. (In Epic 69 ydp, 6} has greater independence.) Except in 6é ye, ye seldom immediately follows a connective. GAA ye, ai ye, xairot ye, pévro. ye are all either rare or un- known in classical Greek. Normally ye either precedes the con- nective (ye wiv, ye pévror, ye wey oy) or follows at an interval {(xai piv... ye, ddA pévror . . . ye). The truth seems to be that ye even in combinations, demands a firmer point d'appui than a mere particle can give. (ii) Preparatory yév and re take precedence of a connective : clearness is gained by placing these particles immediately after the word (or the first word of the group) to which they refer. Zuxpdrns piv yap ...: of re yap ev rij woder Ovres ss (iii) Preparatory pév and re also take precedence of ad- * For the position of roiapuir second in sentence in Hippocrates and in post-classical Greek, see V1.3 below, and p. 567. INTRODUCTION \xi verbial © Wicles. Thus piv 84, pév ye, re 84. (But ye pév seems occasionally to be used for yév ye in Epic and Elegiac, See p. 388.) (iv) Precedence between two combined adverbial particles. ye takes precedence of other adverbial particles, as in yodr, yé rot, ye 84, (The rare reverse order in 84 ye is due to special considerations. See IV above.) ody almost always takes pre- cedence of 8% where neither particle is connective: thus yép ody 84, EAN’ ob 84, wav ody 84. (yey 84 obv is very rare.) Affirma- tive pév takes precedence of ro: and becomes pévrot. (But roe ousts pév in Homeric # rot pév.) 34 ror is almost always preferred to ror 6%, except in obra: 89. mov, a modest particle, readily yields place: yé mou, dimou. VI. THE STYLISTIC IMPORTANCE OF PARTICLES. In the preceding pages I have been concerned with particles mainly from the point of view of grammar and logic, with their origins, natures, and functions, In what follows I shall consider their stylistic features and their distribution over the field of Greek literature, taking into account the distinctions which spring from differences of period, differences of genre, and the individual pre- ferences of various writers. This isa study of great interest and importance. It helps us to appreciate the colour of various styles, to which the use of particles contributes in no slight degree, It may also be of occasional value in establishing the text of a pass- age, and perhaps even in determining, within broad limits, the date and authorship of a work.’ For the most part, this aspect of the particles has been neglected. Writers on the subject have * On the whole it must be confessed that the harvest is disappointingly meagre. Particles do not help us to date Sophocles’ plays, or to determint whether Euripides wrote the A/esws, On the other hand, as 1 hope to hav shown below, they point to certain conclusions regarding the Premetitens and in the dating of Plato’s dialogues they have played a not unimportar part. Here the circumstances are exceptionally favourable. We have in ou hands almost the whole of the very considerable output, extending over a ton life, of an author whose use of particles varies markedly in different worksaxa. AW LKODUL ‘e largely ignored these distinctions ex€ept where they are very striking, though Kiihner is superior to his predecessors in this respect, and the specialized studies provide more information than the general works. (1) Repetition of particles. Before discussing differences of period and so forth, it will be well to consider the general Greek practice with regard to the repetition of particles at a short inter- val. The Greeks seem to have felt about the repetition of words in general that, while artistic repetition is stylistically effective, accidental repetition is not a thing to be sedulously and artificially avoided. (Their attitude to assonance was precisely the same.) The exactness of the significance of Greck pronouns, it is true, often makes repetition of nouns unnecessary. But where repeti- tion is the most convenient course, the Greeks do not boggle at it, and their writings are mostly free from the pitiful periphrases by which some of our own authors have sought to avoid calling a spade a spade more than once. (75 mpocipnuévov is, happily, a good deal rarer in Greek than its English counterparts.) The Greeks felt the same about the repetition of particles. When it is convenient to use the same particle two or three times at short intervals, the same particle is used two or three times (though, when undergraduates write Greek prose, they will cut themselves with knives rather than do this). Thus we find accumulations of ydp (Hdt.i 160.2,199.4: Ant.v 86-7: PLAp.30C,40a: The, 155D: Hyp.£pit.18-19: Arist.Pol.1261a24-6): ye (Hom.E2358: IT30: and see p. 144): d€ (E.£4.73-5: 7745-52: Pherecyd.Fr.18a,105 (8é and ai): Hdt.i 216.3~4): E.Ba.965-6 has & thrice in two lines (but see Murray's app. crit, and Dodds's note): obdé, connective and adverbial (PI.C/it.4o8a): adverbial caf PlR4g5c: X.Gnv 4.42: D.xxxv 50): ofv (Hdt.v 82.1-2): yotw (PILR5548): mou (PL. Cra.sogb): te... wad (Hdt.ix 31.3-5: Pl-Phd.82B-C,108A,1108: Ti.gok), In Hom.g15', S.7r.1151, ddd comes twice in a line: twice in successive lines, S.Aj.852-3, ELB81-2, In Ar.Th.a74-5, P1648-9, two consecutive lines are introduced by roy (cf. Nx. 254-5). In Pux8z0-1 tuayé roe immediately follows tywyé ra. * As will often appear in the course of this book, scholars have not infre- quently introduced by emendation usuges of particles which contlict with the practice of their author, as far as we can know it, map Rane i sR @ INTRODUCTION xiit That I draw largely on Herodotus for examples suggests that there is often a certain naiveté in the repetition: but other of my examples are from more formal and self-conscious prose. Through forgetfulness of this Greek tolerance of repetition, the text has sometimes been needlessly suspected. See $.P4.757-62 (éjra four times in six lines), with Jebb’s excellent note! on this passage and on the threefold dAAd in 645-51.2 It is equally mis- taken to see design in repetitions which are really fortuitous. Thus, van Leeuwen, on ArZys.¥48, ‘Ipsam dein particulam illudens repetit Lysistrata’, I do not think the repetition is intentional here, or in Ar.74.274-5. But I do not mean to deny that there are places where a character in a play throws another character's particle back at him. Thus in S.OT 1005 the Messenger, with a touch of the pawkiness which characterizes Sophoclean messengers, retorts rather impudently with cai pjv, as Xanthias does, most effectively, in Ar.Ra.g26 with of rf ov (Surely you can’t mean...?"). Cf. 6 84 E.£l.236-7. In Ar.Lys. 902 rovydp, fv Soxf echoes the same words in the preceding line {that rocydp belongs to the grand style adds to the joke), just as in Pl.ya9 obxodv éxeivis ety’ éyd echoes g18. In all these cases the repetition has a mocking tone. Cf. also (with more serious inten- tion) A.£u.727-9 (pp. $40-1): S.O7'549-51. In S.PA.854-5 the repeated ro: gives urgency to the appeal. The natural frequency with which common particles recur is not, then, artificially limited by the Greek writers, but is regarded as unobjectionable. In certain cases repetition, while natural or even unavoidable in the context, gives positive gain, I have discussed above (II.2) the effect produced by xaf and otéé in polysyndeton. Further, in addition to the natural tendency of common words to recur, it is probably true that a word will run in a writer's head at a particular time. Like other words,’ particles have, as ‘I doubt, however, whether the iteration, if a shade careless, demands much ‘palliation’, * See also Pearson’s Index to Jebb’s Sophocles, and Radecmacher in Wien, Stud, xivi (1929), pp. 130-2. * For example, Mr. D. L. Page points out to me that dipup occurs four ume in the Trachiniae, nowhere else in Sophocles. «reas in tragedy is confined t A.Supp. (959(?},974,994). The frequency of m\vjdor in the Hes sue and mopta in /7 istargely explained by the subject-matter. a8 eIxiv INTRODUP@ION Dr. Chapman remarks, ‘a certain gregarious tendency’, He points out that in Isoc.xii rofvey occurs nine times in §§ 42-102, not at all in §§ 103-272: in xv, twelve times in §§ 30-124, not at all in §§ 122-204 I note that out of sixteen examples of rofvuy in Herodotus, three are in vii§o.2-4. Comparative Sore is commoner in Trackiniae‘and Bacchae than in other plays. Most of the Sophoclean examples of ov are in the Ajax. ¢g" 6 re is particularly common in (D.} Ivi?' All the examples of roryép in comedy are in the Lysistrata. xa re is especially common in the Hymn to Aphrodite % pry occurs fifteen times in the Parmenides, as often as in all the remaining Platonic dialogues put together. (2) Variations in the employment of particles in different periods, dialects,and styles, and by different authors. Here, owing to the loss of so much Greek literature, we are on slippery ground, For instance, fourth-century tragedy and comedy have vanished, ‘with the exception of meagre fragments. When we find a fourth- century prose usage to be absent from filth-century tragedy or comedy, or from both, can we assume that it was known to the fifth century, but deliberately avoided by the tragedians as alien to the tragic style, or avoided both in tragedy and in comedy as inappropriate to verse? Is it not equally possible that the usage is a late comer, that Aeschylus and Sophocles would have employed it had they known it, and that Moschion, say, actually did employ it? Contrariwise, we have very little Attic prose earlier than 400 B,C. (It is true that Aristophanes does some- thing to make good the paucity of prose in the last quarter of the fifth century. Where, as often, we find his practice agreeing with that of Plato and Xenophon, we may usually safely regard it as colloquial practice.) Again, when we are tempted to talk of individual preferences, may we not be mistaking the character- istics of a type for personal characteristics ? With the ten orators and, to a smaller degree, with the three tragedians, the risk of confusion is less serious. But other types of composition are not so well represented. The extant historians and philosophers, for * In cases like these, where the authorship of a work is unknown, the recurrence of a particle may be due either to ‘gregariousness’ or ta individual preference. m™ INTRODUCTION Ixv example, form far less homogeneous groups. Certainly our path is beset with dangers here, and many of the distinctions which I shall draw are highly speculative. Still, it seemed worth while to include all the material which might appear to point to various conclusions, even at the risk of having a certain amount of chaff mixed up with the grain. (3) Chronological differences. In Homeric Greek particles of emphasis (y¢, 6%, 9, wr), and certain other particles, such as dpa and ‘Epic’ re, are heaped on in almost reckless profusion, and with but little definiteness of application. As the language develops, it tends, in the first place, towards an exacter delimita- tion of the functions of particles (especiatly in prose). Thus emphatic 8% comes to be mainly used in association with certain types of word, and dpa, from denoting interest in general, be- comes specifically an expression of enlightenment or disillusion- ment. yév no longer simply expresses emphasis, but emphasis as an element in contrast. mep loses its independence, and sur- vives only in an ancillary capacity. The use of apodotic had is confined within narrower limits. This is not, of course, to say that delimitation is everywhere exact, There is frequent overlapping, the same idea! being oiten expressed by several different particles or combinations. Expres- sion thus loses in clarity, but gains correspondingly in vari since it is thereby possible, for example, to ring the changes <: GAX4 priv, wai piv, roirvv, in an enumerative series. The crystallizing process is especially discernible in the cis: of combinations. Particles are, for the most part, mo ke: ,. simply piled on one another without regard for redundance. 7 $f and 7 8) udv hardly survive in post-Homeric Greek: $ ra. Gradually yors out of use. Contrariwise, certain combina: become stereotyped, and in some cases (e.g. piv ode, were Meanings irreconcilable with the current usage of their cou Parts, kai develops an adversative sense in xa‘rot (not ye: in Homer) and xai poi, . Tapoke above uf the exacter deimitation of the 1 wg ematolog ally speaking, 1 suppose it is amaccurate to + ds OF word-combinations can mean the hing. 8 believe that the (reeks felt any essential «! Teme wiv in many contents, OF Detneen jes sinIxvi INTRODU' ‘ON some particles in post-Homerie Greek. The functions of others become more diversified. Thus ody, which in Homer is strictly confined to certain associations, with émeé, as, etc., subsequently widens the range of its activities, as rovydp also does. In particu- lar, certain emphatic particles develop a connective sense. (See It above.) The authors who lie between the Epic and Attic periods are unfortunately represented for the most part by scanty fragments. With the dawn of Attic literature in the early fifth century the thread of development can be picked up again, and we can often trace the adolescence or obsolescence of a usage in the course of the fifth and fourth centuries. We can trace, for example, the gradual growth of connective 4 right up to the end of the fourth century, when this use predominates over all others. The case of ov is instructive, Transitional név ov occurs but once in the Wiad, five times in the Odyssey. In these passages odv has the backward reference (‘as I have described, or implied’) which normally accompanies it in the two epics. But a connective force is already beginning to appear. In the Hymns piv obv is propor- tionately commoner (four examples), and the backward reference begins to disappear. We are on the threshold of a new stage, where ody is a full connective. In Aeschylus, ody is fully established as a connecting particle, but is almost confined to questions, a restriction later removed. A similar restriction applies to ovxouy, first found as a connective in statements in E.Med.890. ovxodv in statements is another late development, and Demosthenes and Aeschines are the first writers to use it with any freedom. sv is, on the whole, relatively a late-comer in the field of Attic literature. In general, its frequency increases in the later works of Plato, Lysias, Isocrates, and Xenophon. The common use of xai py in introducing a new character on the stage is hardly yet to be found in Aeschylus. Adversative cat wiv is not found in Pindar, and in Aeschylus it only appears in Agamemnon and Prometheus Vinctus. While dad piv and al piv can be seen coming into use,dAAé piv df and xai pév 84 can be seen passing out of use, The latter are almost confined to Plato, Xenophon, and the earlier orators! (Antiphon, Andocides, Lysias, * The appearance of xai wiv 8) in (D.]Ixi 13 is perhaps of some signi- ficance for determining the date of this composition. INTRODUCTION xvii and Isocrates). The assentient force of wiv oiv, as distinct from the corrective, is absent from fifth-century literature. The formulae of assent, mdvu piv obv, xouidf piv ov, etc., of which Plato was go fond, are common in the Plutus (388 B.C.): see especially 833-8 of that play, where xowidf uty oby is clearly made fun of. Similarly, assentient ydp is rarely found before the fourth century. Its use with an echoed word, common in Plato and in Xenophon’s Socratic works, is clearly parodied, as a new fashion, in Ar.Ec. 773-6. Single re and corresponsive re... re® (excluding ire... efreand olre. .. ofre) grow rarer, on the whole, during the fourth century. As further instances of chronological development, we may notice the replacement of rorydpro: by roryapody and of dAde ydp by GdAa ... ydp, the increased tendency in the fourth century to add ye to apodotic dAAd, and the appearance in the Homeric Hymns of corresponsive xai... xai, hardly found in Hliad and Odyssey. In certain cases we can detect a difference in an author's use of particles between his earticr and his later works. Thus juxtaposed re xaf gets progressively rarer in Andocides. Per- haps, as Fuhr suggests, Andocides gradually adapted himself to oratorical usage in this respect (p. 512). 8) dv and dv 8} are rarer in the later books of Herodotus, which are held by some to have been composed first, than in the earlier. These are but isolated phenomena: but Aeschylus and Plato afford evidence of a more general and more significant character. Aeschylus has left us only seven plays: but, with one excep- tion, they can be dated exactly, or almost exactly, while the dates of Sophocles’ and Euripides’ plays are usually quite un- certain. The Supplices is beyond doubt very early. The Persae was produced in 4: the Septem in 467, and the Trilogy in 458. There remains only the Prometheus Vinctus, which has been assigned by some scholars to about 470, by others to the last 1s perhaps an acy suas * The appearance of misv piv ody in Epich. Fray against the genuineness of this fragment, which Diels as Brounds to the fourth century. ae a8 difficult to say how far the varying frequency ot v4 ,. eis to be attributed to difference of period, how far to ditference of style, (See (5) ad Jin, below.)