Jump to ContentJump to Main Navigation
The Languages of AristophanesAspects of Linguistic Variation in Classical Attic Greek$

Andreas Willi

Print publication date: 2007

Print ISBN-13: 9780199215102

Published to Oxford Scholarship Online: January 2010

DOI: 10.1093/acprof:oso/9780199215102.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 27 February 2017

(p.232) Appendix Aristophanes’ Attic: A Grammatical Sketch

(p.232) Appendix Aristophanes’ Attic: A Grammatical Sketch

The Languages of Aristophanes
Oxford University Press

1. Preliminary Remarks

An exhaustive Aristophanic grammar remains a scholarly desideratum. The following pages cannot fill this gap. Their aim is to present those linguistic features which are of particular diagnostic value in deciding for Aristophanes’ language as a whole—but especially in the stylistically unmarked passages—(1) how conservative or innovative, and (2) how literary (‘artificial’) or conversational (‘realistic’) it is. In other words, this summary treats only those grammatical areas in which Aristophanes’ usage does not correspond to that of other classical Attic writers or, more frequently, where a normative grammar of classical Attic Greek (as represented in our standard handbooks) allows several possible ways of expression to stand alongside each other.

Such parallelisms of two or more competing forms (e.g. older and younger variants) are of particular diagnostic interest and demand frequent statistical references. When neither of the two forms is stylistically marked (i.e. when the choice is determined by criteria such as metrical need or euphony), we may speak of ‘polymorphy’. The rarer form is called the ‘recessive variant’.1 Much of the statistical evidence presented in this appendix has been collected by others, who were working on specific grammatical phenomena. Their data are here supplemented and integrated into a comprehensive picture of Aristophanic grammar.

We must not expect spectacular results from such a collection of material. It largely confirms a view which has been held since antiquity: Aristophanes writes in an extremely pure and somewhat conservative form of Attic Greek.2 Nevertheless, it is worthwhile to point out on what factual basis such intuitive notions as Aristophanic purity and conservatism rest.

Pragmatic or interactional features (e.g. the use of ellipses, anacolutha, deixis, interjections, etc.) and the peculiarities of Aristophanes’ lexicon (figurative language, obscenity, vulgarism, diminutives, comic word formation, etc.), which contribute much to the colloquial appearance of Aristophanic language, have been left aside almost completely since the (p.233) inclusion of such features would not have been of great value for the intended diagnosis; many of them occur mainly or exclusively in comedy and their use can therefore not be compared to that in other texts.3 Even in the three traditional areas of grammar (phonology/phonetics, morphology, syntax), the treatment is necessarily selective.4 However, it is unlikely that a different range of material would have led to divergent general conclusions.

2. Phonology and Phonetics

2.1. Vocalism

2.7.1. Long α

It is a characteristic feature of the Ionic-Attic dialect group that original long α is changed into η. In Attic this change does not take place after ε, ι, ρ Long α in other positions occurs in Aristophanic comedy:

  1. (1) as a product of vowel contraction or compensatory lengthening (e.g. ὀστᾶ, πᾶσα),5

  2. (2) in some analogical formations (aorists in -αυα like ἳσχνανα after aorists of verbs in -ιαίνω/-ραίνω),

  3. (3) in a few word-formation suffixes (bird names in -ᾶς, colloquial -αξ: e.g. φέναξ ‘quack’),

  4. (4) in some colloquial words which may have a non-Attic origin (e.g. γεννάδας ‘noble’, κοάλεμος ‘blockhead’, κόβαλος ‘imp’, σκίταλος ‘saucy’),

  5. (5) with speakers of dialects other than Attic,6

  6. (6) in quotations (e.g. Nub. 30; Vesp. 753; Av. 1337),

  7. (7) in parodies of other literary genres (e.g. tragic lyrics: Thesm. 101–29; Ran. 1264–77, 1331–63),

  8. (p.234)
  9. (8) without consistency in some lyric passages of particular solemnity (often in hymnic songs performed by non-human choruses: e.g. Av. 1058–71; Ran. 209–49).

2.1.2. Compensatory lengthening

In some non-Attic dialects (e.g. East Ionic), the loss of F in the groups vF, ρF and λF was ‘compensated’ by lengthening the vowel of the preceding syllable. In tragedy forms like ξεΐνος and κούρη are used as recessive variants, but they are not allowed into comedy except for parodic purposes (e.g. Thesm. 101 κούραι ‘maidens’, but cf. Thesm. 115 κόραν).

Note. The evidence of Mycenaean Greek e-ne-ka (without -w-) suggests that ἕνεκα/εἴνεκα does not go back to *ἕvFεκα and that εἵνεκα represents an artificial literary variant with metrical lengthening that made its way into non-epic Ionic. Nevertheless, the shape of this preposition is best discussed here because the case is synchronically similar to that of compensatory lengthening. By far the most common Aristophanic form of the preposition is οὕνεκα, which originated as a sandhi variant when cases like δήμου (ἕ)vεκα or δημουνεκα were reanalysed as δήμου οὕνεκα. Unlike ἕνεκα, οὕνεκα is in fact regularly used as a postposition.7 Beside οὕνεκα, the original ἕνεκα —a convenient metrical alternative—is well-attested. Apart from these two, the Aristophanic manuscripts sometimes have εἵνεκα. Guided by Wackernagel’s opinion that εἵνεκα is a ghost-form in Attic,8 editors often replace εἵνεκα with the metrically equivalent οὕνεκα. But although Wackernagel may have been right, one must not argue that Aristophanes or Demosthenes should also use ξεῖvos and μούνος if they use εἵνεκα. At a time when Attic was sensibly influenced by Ionic, a sandhi product δημουνεκα could also be reanalysed as δήμον εἵνεκα. If our texts must be regularized, Wackernagel’s view is the only possible one, but there is a slight chance that it obscures real variation in late fifth-century Attic Greek.9

2.1.3. εἰς υς. ἐς

Word-final -vs lost its nasal at an early stage, without trace if the next word began with a consonant but triggering another (earlier) process of compensatory lengthening if the next word began with a vowel. So an original *ἐvs would develop into εἰς and ἐς respectively. The former was generalized in Attic, the latter in Ionic. Whereas tragedy uses ἐς as a polymorphic variant form, comedy has only εἰς (except in fossilized idioms such as ἐς κóρακας (p.235) ‘be hanged!’). other comic occurrences of ἐς (of which only the prevocalic cases are significant) are paratragic or non-Attic.10

2.1.4. Vowel contraction

Attic Greek eliminated word-internal hiatus more consistently than other dialects, most notably also in groups like εо and εω (though not in bisyllabic words or when a digamma had separated the two vowels). Uncontracted vocalism in comedy usually11 indicates parody of tragic, lyric, and epic poetry,12 or it serves as a literary embellishment in anapaests and lyric metres.13 Thus, the uncontracted suffix -óεις appears only in a few lyric, epic, or para-epic passages,14 whereas the contracted variant -ούς is commonly used in names for cakes and pastry (μελιτούς ‘honey cake’, πλακούς ‘flat cake’, πυραμούς ‘wheat cake’, etc.).

Stylistically unmarked polymorphy exists in a few particular cases.

  1. (1) The name of the goddess Athena occurs both as Ἀθηναία (Pax 271; Av. 828, 1653) and as Ἀθηνᾶ (Pax 218) in Aristophanic trimeters. In the Attic inscriptions, too, Άθηναία continues to appear, even in private dedications, down to the fourth century.15

  2. (2) Names in -κλής also occur in the uncontracted form -κλέης. The nominatives Пερικλέης, Σοφοκλέης, ίεροκλέης, and мεγακλέης Stand against Θεμιςτοκλής and Ηρακλής, and the dative Ηρακλέει (Av. 567, anap.) differs from Άνδροκλεΐ (vesp. 1187).16 The uncontracted forms do not necessarily reflect an earlier state since they may have been recreated in analogy with the genitive in -κλέoVS (which is not further contracted).17

2.1.5. ἐάν vs. ἢν

It is sometimes claimed that the conditional conjunction ἢν for ἐάν is Ionic rather than Attic.18 In reality, it is the regularly contracted Attic form. Whereas ἐάν is usual in the inscriptions,19 it is the recessive variant in Aristophanes, who has c.70 ἐάν and c.270 ἢν The crasis form κἂν (= καὶ ἐάv, (p.236) c.70 examples) must not be used as evidence for a third variant ἂν, although this is occasionally found in the inscriptions.20

2.1.6. ἑαυτ-/σεαυτ- vs αὑτ-/σαυτ-

The short variant αύτ-/σαυτ- of the reflexive pronoun reflects an allegro pronunciation rather than a regular contraction. In the eleven complete plays I have counted the following figures (the numbers for trimeters only are in brackets): ἑαυτ - 23 (6); αύτ- 44 (23); ςεαυτ - 25 (16); σαυτ- 71 (44). obviously the long forms are the recessive variants. The small number of ἑavr- in trimeters suggests that ἑαυτ- may not have been heard very often in ordinary speech. In anapaests ἑαυτ- and αὑτ- are distributed equally (11 : 12), but ἑαυτ - occurs 7 times at the ‘dactylic’ line-end where divergences from real-life language are most common. The comparison of Aristophanes’ early and later comedies does not provide evidence for a gradual replacement of the longer variants by the shorter forms. The Attic inscriptions prefer ἑαυτ- throughout the fifth and fourth centuries.21 Surprisingly, tragedy (especially Euripides) favours the shorter variants more clearly than (Aristophanic) comedy.22 This may be a sign that Aristophanes is phonologically conservative.23

2.1.7. ποιέω vs. ποέω, etc.

Prevocalic αι, ει, and οι often lose their ι in classical Attic orthography (even when originally the ι was followed by *F). Presumably the second element of these diphthongs was not always pronounced, even if it was written. This is certain when the quantity of the ‘diphthong’ syllable is affected (but not in pairs like κλαίω/κλάω where the first syllable is always long24). The most frequent case in comedy is the parallelism of ποιέω and ποέω25 How to spell this verb in modern editions is a secondary question, but the alternation between a long and a short first syllable is a linguistic fact. Taking Acharnians, Knights, and Clouds as a sample, I have counted the following figures: πο(ι)- long, 21; πο(ι)- short, 58; πο(ι)- ambiguous, 3. Given that 33 out of 39 unambiguous occurrences in iambic trimeters are short, long πo(ι)- must be regarded as an artificial conservatism for metrical (p.237) convenience. Despite the epigraphic convention of writing πο- only before ε and η,26 the following vowel does not seem to influence the length of the syllable.

2.2. Consonantism

2.2.1. σσ vs.ττ

In Attic (as in Boeotian and West Ionic), original groups like *κy, *χy, *τF, and sometimes *τy and *θy develop into ττ, not σσ (as for instance in East Ionic).27 Unlike tragedy, comedy adopts Attic ττ. Some scholars have wondered whether σσ and ττ were just spelling variants for the same sound. This is not only unlikely28 but virtually excluded by the manuscripts of comedy where σσ does appear in the same well-defined environments as long α: in words of foreign stock (e.g. νάρκισσος ‘narcissus’), some place names (пαρνασσός), solemn hymns,29 foreign dialects,30 parodies or quotations of epic and lyric poetry,31 paratragedy, and tragic quotations.32 If there had not been a difference in pronunciation, it would be difficult to explain why σσ and ττ should have been distinguished orthographically.

2.2.2. ρσ vs. ρρ

Mutatis mutandis the preceding remarks on ττ are also valid for Attic ρρ from original ρσ (which was retained in Ionic and other dialects but in Attic only under analogical pressure, e.g. in the dative plural of r-stems). Because this group is infrequent anyway, comic examples of non-Attic ρα are extremely rare. They do occur in the foreign word βύρσα ‘leather’ and its derivatives as well as in the parody of Agathon’s lyrics (Thesm. 125).

2.2.3. ξυv(-) vs. συν(-)

In the Attic inscriptions of the late fifth century, the preposition or preverb συν(-) predominates over ξυv (-). Similarly συν (-) is about three or four times as frequent as its older variant ξυν(-) in those Aristophanic passages in which the metre allows a decision.33 As a preposition ξύν/σύν (instead of μετά) occurs only in the same elevated contexts as σσ or long α34 and in a few (p.238) idiomatic expressions (ξύν ὅπλοις ‘under arms’, ξὐν θεoί ς ‘with the gods’, ξὺν νῷ ‘wisely’). Here the old form may have been fossilized so that the coexistence of συν(-) and ξυν(-) is not a case of true polymorphy with two exchangeable variants.

2.3. Suprasegmental Features

2.3.1. Accent

The accentuation of our texts was introduced long after the time of Aristophanes. Hence, little can be said about how regular or irregular comic pronunciation was in this respect. The sneering reference at Ran. 303–4 to the tragic actor Hegelochus’ slip of saying γαλῆν ‘weasel’ instead of γαλήν’ ‘calm (of the sea)’ suggests, however, that the traditional pitch accent (and not the stress accent that came to be used in later Greek) was still the norm in both comic and tragic speech.

2.3.2. Correptio Attica

A short-vowel syllable before most groups of plosive + liquid or nasal (i.e. muta cum liquida) is short in Attic (correptio Attica). In lyric metres and in anapaests non-observance of correptio Attica is quite common.35Correption is usual in both tragic and comic trimeters (76.5% observance in Aeschylus, 69.7% in Sophocles, 69.2% in Euripides), but comedy is stricter (86.5% observance in Aristophanes’ trimeters and tetrameters).36 Since some groups (βλ, γλ, γν, γμ, δν, δμ) regularly break the rule and should therefore be disregarded, the real figures may be even clearer. There is only a handful of non-parodic comic trimeters in which correptio Attica is not observed.37

2.3.3. Avoiding hiatus: crasis, elision, prodelision

With crasis, elision, and prodelision, comedy is more tolerant than tragedy. Presumably this reflects the usage of spoken Attic.38 In comedy, unlike tragedy, verbal forms in -μαι, -σαι, -ται, -σθαι, and –ναι are freely elided. Among the more spectacular cases of crasis, which involve two polysyllabic words, one may note for instance ἐγᾦδα (= ἐγὼ οἷδα, e.g. Ach. (p.239) 904), περιόψομἀπελΘόντ’ (= περιόψομαι ἀπελθόντ Ran. 509)> or μέντοϋφαςκεν (=μέντοτ ἓφαςκεν, Eccl. 410). I am not aware of any comparative statistics.39

According to Platnauer, there are 2.3 cases of prodelision in every 100 lines of Aristophanes (as against 0.4 in Aeschylus, 1.3 in Sophocles, and 0.6 in Euripides).40 Platnauer observes that prodelision becomes rarer in Aristophanes’ last plays (and even more so in Menander).

While hiatus is mostly avoided, it is tolerated (or rather created) in οὐδὲ/μηδὲ εἷς, where it strengthens the expressive force of this variant for οὐδείς/μηδείς. The gradual spread of οὐδὲ εἷς is reflected in Aristophanic comedy: the hiatus form first occurs in Lys. 1045 and Ran. 927, but it is common only in the last play (Plut. 37, 138, 1115, 1182).41

2.3.4. v ephelkystikon

With certain inflectional forms, most notably the dative plural in -(οι/αι)σι, the third plural in -ουςι, ἐςτί/εἰσί, and the aorist and imperfect third singular in -ε, hiatus can be avoided not only by elision but also by the addition of a v ephelkystikon. This device, in Aristophanes apparently more common in anapaestic metres than in iambic trimeters, is increasingly frequent in fourth-century Attic.42 If we take the occurrences of ἐστί vs. ἐστίν according to Todd’s Aristophanic index as one sample and the first 500 lines of Acharnians as another, it seems that forms with and without v ephelkystikon are distributed almost equally, though perhaps with a slight predominance of the shorter forms.

2.4. Varia

2.4.1. Phonological uncertainties

With some of the phenomena discussed here one has to rely entirely on the transmitted orthography. As long as two variants are well differentiated and the transmission does not contradict our expectations, this procedure is relatively safe (cf. e.g. σσ vs. ττ).

In other cases we have to admit our ignorance. Take for instance the ending -ηι (-η) of the a-stem dative singular and of the second-person middle. The Attic inscriptions show some attestations of an α-stem dative (p.240) in -ει as early as the fifth and fourth centuries. This may suggest that some people pronounced this ending as [ei] (or [e:]?), but we do not know whether most of them did. Somewhat inconsistently modern editors often prefer -ει for middle forms such as <βούλει> ‘you want’ (or the dative <πόλει> rather than <πόληι> although the latter is the regular orthography of the inscriptions before c.370 BC) but not for <κόρει> ‘to the girl’.43

Similarly, at the time of old Comedy the standard ending for the nominative plural of words in -εύς was ῆς if the inscriptional spellings are taken as a guideline.44 But how can we know that this was not actually pronounced in the same way as the later –εῖ ς?

A comparable problem of consonantism is the ‘correct’ spelling of words beginning with either σφ-/σχ- or σπ-/σκ- (e.g. σπόγγος ‘sponge’, σκινδάλαμος ‘splinter’). Even if the orthographic question could be solved,45 it would tell us nothing about the phonetic realization.

2.4.2. Puns as evidence

on the basis of cumulative evidence from a group of comic puns which are all supposed to work with the same kind of assonances, Perpillou has argued for a widespread pronunciation of αι as [e] or [e:]. However, his material is hardly sufficient to prove the suggestion,46 and it seems that this method of reconstructing the phonetics of the comic stage is not practicable given the low level of phonetic similarity that informs much of Aristophanes’ punning.47

2.5. Conclusion: Phonology and Phonetics

Where it is possible to make a judgement, the phonetic and phonological level of Aristophanic grammar corresponds well with the classical Attic system. Deviations are nearly always introduced for precise stylistic reasons (e.g. in lyric parts and literary parody). Polymorphy with two stylistically equivalent variants occurs only in a few marginal cases (ἐáv/ἥv, ἑαυτ-/αὑτ-, ποιέω/ποέω), and quite possibly both variants were used alongside each other even in non-literary Attic. The case of ἑαυτ-/αὑτ- (and that of (p.241) Άθηναία) may point to a certain phonetic conservatism (or aim for careful speech), but artificial archaizing is absent (cf. ξυν-/συν·). only with ἑavτ-/αὑτ- does tragedy seem to be statistically closer to what one might expect in ‘real-life’ Attic (contrast e.g. σσ/ττ, correptio Attica).

3. Morphology

3.1. Nouns

3.1.1. Dative plural of o-stems

Throughout the fifth century the Attic inscriptions show infrequent examples of the long dative plural ending оιςι(ν) rather than the usual -оις. The last securely dated examples of -οιςι(ν) in public documents appear around 420 BC, but because official inscriptions are often conservative it is safe to assume that the long form had disappeared from spoken Attic long before.48 In Aristophanes -оις is far more common already in the earlier plays, but -οιςι(ν) does exist, also in the late plays, as a recessive variant. In Acharnians, Knights, and Clouds, I have counted 271 -оις against 104 -οιςι(ν). Even if the 88 cases of prevocalic -оις are disregarded (since they might be interpreted as -οις’), - оις is twice as strong as -οιςι(ν). Nevertheless, it is a revealing sign of morphological conservatism in Aristophanes if an old, and probably lost, form like - οιςι(ν) has such a strong position. Metrical reasons alone are no sufficient explanation since Menander manages to write his plays virtually without - οιςι(ν).49 There is no significant difference between the treatment of the dative plural endings in nouns and pronouns.

3.1.2. Dative plural of a-stems

Although the situation is not exactly the same, the Aristophanic treatment of a-stem and o-stem dative plurals is more or less identical. The original ending of the α-stem dative plural was -ησι (or -ασι after ε, ι, ρ). This ending is regular in the inscriptions until around 420 BC, when it is quickly replaced by -αις.50 In Acharnians, Knights, and Clouds there are 122 forms in - αις (of which 36 are prevocalic) and 39 in -αισι(ν). Where the longer form occurs, we might wonder whether Aristophanes wrote -αισι(ν), as transmitted, or -ησι(ν), as on the inscriptions (where -αισι is very rare). on the whole, the case for -αισι(ν) seems stronger: with -ησι(ν) we would have to assume a somewhat gratuitous extensive orthographical revision of the text. In any case the strength of -αισι(ν) in Aristophanes reflects literary (p.242) convention rather than the real frequency of the long endings in late fifth-century spoken Attic.

3.1.3. Attic declension

The so-called Attic declension (e.g. in vεώς, λεώς, ἵλεως), which disappeared in Koine Greek, is alive and well in Aristophanes. Forms like ναός or λαός are stylistically conditioned.51 Mε ν έλαoς once occurs at the hexameter-like end of an anapaest (Av. 509).

3.1.4. Irregular declension forms

In Aristophanes, as elsewhere in classical Attic, the word νίός has an o-stem nominative and accusative singular, but u-stem forms in the other cases.52Similarly, forms such as δέvδρεσι (Av. 1066), κλάδεσι (Av. 239), and κρίνεσι (Nub. 911) are ‘regular’ within their heteroclitic declension patterns.

Irregular declension forms of other stems either have a distinctive stylistic function53 or belong to marginal and (synchronically) irregular words where people may have been uncertain about the ‘correct’ forms.54 The only truly surprising metrical licence is the stylistically unmarked genitive θυγατέρος in a trimeter (Vesp. 1397).

3.2. Adjectives

3.2.1 Comparative stems

In classical Attic the comparatives in -(ί)ων are regular n-stems in most cases: -(ί)ων, -(ί)ονος. only the forms ending in -(ί)ονα (acc. sg. masc./fem., nom./acc. pl. ntr.) and -(ί)ονες/-(ί)ονας (nom./acc. pl. masc./fem.) compete against remnants of the older s-stem declension: -(ί)ω, -(ί)ους. In the inscriptions, probably because of their conservatism, the longer forms spread into the remaining domain of the s-stem forms long after Aristophanic times.55 To judge from a sample of ten of the most frequent comparatives in -(ί)ων,56 Aristophanes uses both variants with the same frequency. The longer forms are slightly predominant with -ονα/-ω (27 : (p.243) 21), the shorter ones with -ονες/-ονας/-ους (4:5: 15). It is just possible that one variant (presumably the older and shorter variant rather than the younger and longer one) was purely literary, but the comic text suggests that both were commonly used.

3.2.2. Comparatives and superlatives in -ίστερος/-ίστατος

Since the comparative and superlative are highly expressive linguistic categories, their intensifying character is frequently strengthened, especially in colloquial language, through peculiar morphological developments. one possibility is the use of the suffix -ίστερος/-ίστατος, often with a pejorative connotation. In words such as ποτίστατος ‘great boozer’ (Thesm. 735) and perhaps even κλεπτίστατος ‘most thievish’ (Plut. 27), which are based on adjectival -πότις and κλέπτης respectively, the suffix may have been regarded as semi-regular. From here it was transferred to some other stems (mostly o-stems): πτωχίστερος ‘more beggarly’ (Ach. 425), μονοφαγίστατος ‘most alone-eating’ (Vesp. 923), μισοπορπακίστατος ‘most shield-band-hating’ (Pax 662), λαλίστερος ‘more talkative’ (Ran. 91, Ar. fr. 684), ψευδίςτατος ‘most lying’ (Ar. fr. 920).57 Where it borders on word formation, comic morphology is somewhat colloquial.

3.2.3. Irregular comparatives and superlatives

There are a few cases of analytic comparatives (Vesp. 1105 μᾶλλον όξύθυμον ‘more sharp-spirited’; Plut. 747 μᾶλλον τυφλόν ‘more blind’).58 other irregular comparatives and superlatives include colloquial intensifications (frequent πρώτιστος ‘firstest’; Eq. 352, Plut. 182 μονώτατος ‘alonest’; Eq. 1165 προτεραίτερος ‘firster’; Vesp. 1502 μέσατος ‘middlest’; Plut. 83 αὐτότατος ‘self-est’), but some of them may have been designed for comic effect like Δαναώτατος ‘Danaosest’ (Ar. fr. 270). There are one or two cases where metrical need played a role: Nub. 790 ἐπιλησμότατος ‘most forgetful’ (cf. Cratinus fr. 451 and Alexis fr. 317, perhaps attesting a positive ἐπίλησμος), Lys. 776 καταπυγωυέστερος ‘more pathic’.59

3.3. Pronouns

3.3.1. Neuter singular

In old Comedy the neuter pronominal forms ταὐτόν, τoιoῦτoν, τοσοῦτον virtually never occur without the final -v: in Aristophanes, Eq. 1234, where the manuscripts have τοσοῦτο, is metrically ambiguous, but Plut. 361 τοιοῦτο (p.244) need not be changed.60 Since forms like τοιούτο appear quite often in Middle (and New) Comedy, they may have been used occasionally by 400 BC. If so, this would be an additional indirect argument for Aristophanes’ morphological conservatism.

3.3.2. Dual of a-stem pronouns

Whereas nominal a-stems regularly have a dual in , pronominal dual forms in (τά, ταύτα) are not attested in the Attic inscriptions of the classical period, and ταῖυ is exceedingly rare.61 It therefore depends on our belief in the transmission whether we allow forms like τά and ταῖν into the Aristophanic text (or replace them by τώ, τοίν). While it would be wrong to postulate τά for frequent collocations such as τὼ θεώ ‘the two goddesses’ or τὼ χεῖρε, it may also be dangerous to excise τά where it is backed by some manuscripts (Eq. 424, 484; also Pax 847 ταύτα?). In the genitive and dative, ταῖν is hardly ever found instead of ταῖν in the transmitted text (also ταίνδε: Eccl. 1106). Analogical pressure may well have led to the creation of such forms in spoken Attic, even if they were not recognized as correct by those who wrote epigraphic texts.62

3.3.3. Tendency towards bisyllabism (τοῦ υs. τίνος, ὃτου vs. οὕτινος, etc.)

In the interrogative pronoun the shorter and older forms τοῦ; (33 exx.) and τῷ; (26) are still more frequent than their polymorphic variants τίνος (17) and τίνι; (4). In the indefinite pronoun, πινός (12) and πιví (II) have already overtaken τov (9) and τῳ (9) (which remain predominant in the inscriptions until around 350 BC63).

However, Aristophanes never uses οὕτινος, ᾥτινι instead of ὃτου, ὃτῳ (οἷστιςι occurs once in a parodie hexameter: Pax 1279). Morphological conservatism may here be coupled with a preference for bisyllabic pronominal forms, which yielded the best equilibrium between the two competing forces of linguistic economy, clarity and effort. The tendency towards bisyllabism also explains the occasional replacement of οἳτινες/αἳτινες by ὃσοι/ὃσαι (Ach. 862; Lys. 268; Eccl. 112).

3.3.4. Deictic -ί, (ό)τιή

Many personal pronouns and certain adverbs are reinforced by the addition of a deictic (όδ ί, ούτοσί, ἐνθαδί, etc.). There are more than 600 examples of deictic - ί in Aristophanes (though very few in the parabasis and (p.245) lyric or paratragic parts),64 whereas it is virtually absent from the official inscriptions, from tragedy, and from Thucydidean prose. The distribution shows that deictic was particularly strong in colloquial Attic but avoided in more formal text genres. According to Dover’s statistics, the ratio between forms with and forms without - ί is 1 : 4 in Aristophanes (contrast e.g. Lysias with 1:17 and Plato with 1 : 30).65

Similarly, comedy often reinforces with an added particle both interrogative causal τί; ‘why?’ and the conjunction ὃτι, mainly in its causal use: τιή;, ὁτιή. Since τιή; occurs already in epic poetry, such ‘strenghtened’ variants must have had a long subliterary history and should not be regarded as recent morphological innovations.66

3.4. VERBS

3.4.1. Endings -μεθα vs. -μεσθα

The most frequent case of verbal polymorphy is that of the first-person plural middle and passive endings -μεθα and -μεσθα. It is very unlikely that -μεσθα was used in spoken Attic, but in Aristophanic comedy it is a common recessive variant (76 -μεσθα : 177 -μεθα). Although the tolerance for -μεσθα seems greater in lyric passages, it is not confined to them. In later comedy, -μεσθα gradually disappears like the dative plural in -οισι/-αισι.67 Optative of contracted verbs

The old set of singular optative endings for contracted verbs (-οῖμι, -οῖς, -oῖ, and, for verbs in -άω, -?μι, etc.) is replaced in Attic by the remodelled set -οίην, -οίης, -οίη (after athematic forms like διδοίην; for verbs in -Άω: -ῴην, etc.). In tragedy the endings of the former set still serve as recessive variants, but in Aristophanes there are only two well-attested examples in stylistically marked contexts (Eq. 1131, lyr.; Pax 1076a/b, para-epic hexameter).68 Aristophanes never uses plural forms with the optative marker -ιη-, not even with athematic verbs or in the aorist passive.69

(p.246) ‘Aeolic’optative

In the active aorist optative the endings for the second and third singular and third plural are either ‘Aeolic’ -(σ)ειασ, -( σ) ειε, -(σ) ειαν or ‘non-Aeolic’ -(σ)αισ, -(σ)αι, -( σ) αιεν. Statistics show that the ‘Aeolic’ forms were regular in fifth-century Attic (Table A.1).70 The Aristophanic manuscripts never have -(σ)αιεν (but 6 times -( σ) ειαv) nor an entirely certain case of -(σ)αι (but 44 times -( σ)ειε).71 Real polymorphy seems to exist only in the second singular (13 -(σ)ειας: 8 -(σ)αις). In Middle and New Comedy, -(σ)αισ takes over completely, but Aristophanes is conservative: 2 out of 8 -(σ)αις in Aristophanes belong to the late Plutus, and 3 of the earlier examples stand in anapaests where -(σ)αις may be regarded as a poetic licence rather than as a colloquial innovation.

TABLE A.1. Frequency of ‘Aeolic’ and ‘non-Aeolic’ optative endings in Attic drama















old Comedy (except Ar.)


Middle/New Comedy


15 Middle optative (and third-plural perf. med.-pass.): -ατο vs. -ντο.

The third-person plural of the middle optative presents a further case of true polymorphy in Aristophanes. In Aeschylus and Sophocles -oιατo/ -αιατο still predominate over -οιντο/-αιντο, but with Euripides and Aristophanes a reversal takes place (Aristophanes: 16 -vτo : 5 -ατo).72 It is impossible to say how often, if at all, the forms in -ατο could still be heard in spoken Attic.

In the medio-passive third-person plural of the perfect and pluperfect, the endings -αται/-ατο are not found in Aristophanes. Instead, periphrastic forms or, in the case of verbal stems ending in a vowel or diphthong, forms in -vται are used.73

(p.247) ἧ vs. ἧν; pluperfect endings; οἷσθα

In the first-singular imperfect of εἰμί, the later form ἧv is guaranteed (since it prevents hiatus) only in Plutus (29, 695, 822). of course, this need not imply that Aristophanes, unlike for instance Euripides, never used ἧv before.74

In the first-singular pluperfect both and -ειv seem to occur: in Av. 511 ᾓδη ί knew’ is certain (against some of the manuscripts), but in vesp. 558, 635, and Pax 1182 the later -ειv stands before a vowel. In the second person -ᾓδησΘα (or -εισΘα) is once guaranteed (Eccl. 551), ᾓδειδ never (though transmitted).75 Unlike Cratinus (fr. 112), Aristophanes does not use οἷσΘασ (nor οἷδασ) instead of οἷσΘα. Imperative endings

With certain athematic verbs such as ἳσταμαι or τίΘεμαι the middle imperative occurs in a contracted (ἳστω, τίΘου, also κΆΘου ‘sit down’) and an uncon-tracted (ἳστασο, τίΘεσο, also κΆΘησο) form. In the latter, the original intervocalic -σ- has been retained or analogically restored. Surprisingly, Aristophanes and Euripides do not share the preference of both early tragedy and comedy for the contracted variants.76 Whether Aristophanes’ treatment was conservative or not, it may show some concern for morphological clarity.

There are no Aristophanic examples of the third-plural active imperative -τωσαν, which replaced -ντων in later Greek.

3.4.2. Augment and reduplication

The augment is omitted in comic dialogue only once in Aristophanes, in the para-epic verb βινεσκόμην ‘I used to fuck’ (Eq. 1242). Elsewhere omission occurs in para-epic hexameters and in some parodie or high-flown lyric passages.77

(p.248) With μέλλω, βούλομαι, and δύναμαι, fourth-century Attic began to use an augment ἠ- on a regular basis, but Aristophanes has only ἢμελλ- (Ran. 1038 and Eccl. 597, against 7 examples of ἒμελλ-), never ἠβουλ- and ἠδυν- (ἠβουλ-and ἐδυν- being common).78 Since at least ἠβουλ-, which is modelled upon ἢΘελ- from ἐΘέλω,79 is found as early as the fifth century, this is another indication of morphological conservatism. With κΆΘημαι, καΘέζομαι, and καΘίζω, initial and interior augment coexist even in trimeters (though ἐκαΘ- is normal).80

The perfect of ὁράω is ἑόρακα, not ἑώρακα (e.g. Thesm. 32–3; Plut. 98). on augmented ἒχρην see Chapter 5.8.2.

3.4.3. VERBAL STEMS Present: thematization of athematic verbs

The thematization of athematic verbs such as (ἀπ)όλλυμι, δείκνυμι, μὄνυμι, which is frequent in Middle Comedy, is found only in the late Aristophanes (Plut. 719 σνμπαραμειγνύων; cf. his older contemporary Pherecrates’ fr. 152. 9 and further instances even in writers like Thucydides and Andocides).81

The only case where (something like) thematization is frequent is the imperative of verbs in -βαίνω with either athematic (aoristic) -βηΘι or quasi -thematic -βα (as if from *-βάω cf. δίδου, τίΘει, ἳει). Among the tragic writers only Euripides uses these forms, which may have been colloquial standard.82 Aorist

From the middle of the fourth century onwards, the Attic inscriptions show the aorist marker -κ- of the reduplicated athematic verbs δίδωμι, τίΘημι, and ‘ἳημι also in the plural (i.e. ἒδωκαν for ἒδοσαν, etc.). Considering that epigraphic conventions are usually conservative, such forms may already have been in widespread use by 400 BC; they are preferred to the older ones in Middle and New Comedy.83 In Aristophanes there are only two cases and one of them reflects the free use of such forms in epic poetry rather than linguistic innovation (Nub. 968 παρέδωκαν at the end of an anapaestic tetrameter, leaving only Ach. 101 ξυνήκαΘ’ as a ‘recent’ form).84

(p.249) In tragedy, presumably under Ionic influence, weak aorists sometimes replace strong aorists in the passive voice. In Aristophanic comedy, Thesm. 1128 στρεφθῶ (paratragic) and Lys. 526 συλλεχθείσαις (anap.) are isolated.85Regular alternation is found only with (ἀπ-/δι-)ηλλά;γην vs. (ἀπ-/δι-)ηλλἀ;χθην, though again with a preference for the strong form (15:5 examples, the latter probably being influenced by the future -αλλαχθήσομαι).

Aristophanes does not yet have aorist forms like εἷπα/ἢνεγκα or εἷπες, which are found in various fourth-century authors.86 At the same time, the literary aorist ἒλεξα beside εἷπov is much rarer in Aristophanes than for instance in Euripides (but not as rare as in later comedy).87 Perfect

The old system of οἷδα/ἲσμεν is stable in Aristophanes, but the infinitive and participle εἰκέναι/εἰκώσ of ἒοικα alternate with the rarer variants ἐοικέvαi/ἐoικώσ.88 The parallelism of ἐστώς, etc. (15 examples, including compound forms) vs. ἑστηκώς (5) seems to be another case of metrically conditioned polymorphy. In the perfect of γίγνομαι, the middle γεγένημαι (28) is already stronger than γέγονα (11).

The later aspirated perfects (such as πέπραχα: cf. e.g. Ran. 302 πεπρἀγαμεν) are virtually absent from Aristophanes (but note Lys. 952 ἐπιτέτριφεν; forms like εΐληφε are of course common, Lys. 1003 ὑποκεκύφαμεσ is etymologically justified, and Ar. fr. 615 λέλαφας ‘you have gulped down’ may contain a stem λαφ- as in λαφύσσω ‘to gulp down’). Future

Contracted futures like ἀμφιῶ (from ἀμφιέννυμι), ἐλῶ, κρεμῶ, πετῶ, or σκεδῶ

tighten rather than loosen their grip on Attic over time. When Aristophanes has βιβῶ (Av. 425, 1570; the Attic writers Andocides and Lysias use βιβασ-) and, despite the ample attestation of κολἀ;σω/-ομαι in classical Attic, even κολῶμαι ί will punish’ (Eq. 456; vesp. 244), this may be a recent feature.89

The use of ‘Doric’ futures like φευξούμενον (e.g. Ach. 1129, but vesp. 157 ἐκφεύξεται), πνευσεῖται (Ran. 1221), or χεσεῖσθαι (vesp. 941) may have been an option in spoken Attic and need not reflect metrical constraints even if it contravenes prose usage.90

(p.250) In analogy to τυπτήσω and παιήσω, the future of βἀλλω is either βαλλήσω (vesp. 222, 1491) or βαλῶ (Ach. 283; compound verbs). Similarly, δόξω competes with occasional δοκήσω.91 The verb ῥίπτω can be treated as a verb in -έω even in the present (vesp. 59; Eccl. 507). Like ἔχω with σχήσω and ἔξω, τρέχω has two futures with δραμοῦμαι (vesp. 138) beside θρέξομαι, and it is not easy to see a semantic distinction between the two.92

3.4.4. Tmesis

The separation of a preverb from the verbal stem (‘tmesis’) is commonly regarded as poetic (cf. Pl. Phdr. 237a). A similar, though possibly independent, phenomenon occurs in comic dialogue, but here the two components are separated only by particles or other enclitic words. The tone of some of the Aristophanic examples (and the frequency with ἐπόλλυμι) suggests that tmesis was a means of colloquial intensification.93one may compare the ‘adverbial tmesis’ of words like ἐvταυθí, which becomes ἐνγεταυθί (Thesm. 646).

3.5. Morphological Uncertainties

As with phonology, there are cases of uncertainty in mrphology. Apart from those mentioned already, they include the accusative singular of s-stem personal names, where the inscriptions suggest that -ην was normal by 400 BC and need not be changed into ,94 the aorist of πέτομαι (both ἐπτόμην and ἐπτἀμην occur in the manuscripts),95 or the controversial ‘ Lautensach’s Law’, according to which compound verbs starting with εὐ-’well, good’ have no temporal augment ηὐ-.96

3.6. Conclusion: Morphology

Formal variation is more frequent in Aristophanes’ morphology than in his phonology. In some cases it is likely that one of the two variant forms was not used in contemporary spoken Attic (-οισι/-αισι, -μεσθα, perhaps -αro). Elsewhere the parallelism of two forms is not ‘artificial’ but points to the existence of synchronic variation also in ‘real’ language. Here, (p.251) Aristophanes does not usually show a clear preference for the younger variants (cf. e.g. comparative stems, τοῦ;/τῷ;, ‘Aeolic’ optatives, augment with μέλλω, first singular ἧν). A conservative attitude is indicated by the absence or rarity of certain features (e.g. aorists like ἔδωκαν, οἶσθασ, thema-tized athematic verbs, weak instead of strong aorists). of course this does not preclude the use of colloquial features since ‘colloquial’ is not always the same as ‘recent’ (cf. deictic -ί, τιή;, tmesis, probably comparatives and superlatives in -ίστερος/-ίστατος). Generally speaking, Aristophanes seems to exploit variation (or synchronic ‘irregularity’) more readily in verbal than in nominal morphology, but obviously the rich verbal system of classical Greek also provided him with more numerous opportunities to do so.

4. Syntax

4.1. Nouns

4.1.1. Case Nominative and vocative

Nub. 1168 and Aυ. 1467 may be interpreted as containing a nominative instead of a vocative, but apparent cases elsewhere must be taken as exclamatory nominatives.97 The switch from a vocative to a nominative in a list of addressees is very old, but it is not a strict rule (contrast Nub. 264–5 with Ach. 55).

The vocative is preceded by ὦ much more frequently than not, but the two variants are essentially equivalent (compare e.g. Ach. 95 and 818, 284 and 1018, 887 and 1174, 1132–3 and 1136–7).98 Accusative

Directional accusatives without a preposition are always stylistically marked in Aristophanes and extremely rare even in parody.99 The same is true for the non-prepositional ablatival genitive100 and local dative101 (p.252) although the latter is fossilized in forms like Mαραθῶυι, Ὀλυμπίασι, κεφαλῆσι (which are comparable to a directional Ἁλιμουντάδε or an ablatival κριῶθεν).

Participial absolute accusatives (6 ἐξóv, 2 μετόν,1 δέον, 1 προςῆκον, 1 εἰρημένον) are relatively rare (cf. Euripides with 16 έξόν/παρόν). Adverbial τυχόν (ἴςως) ‘perhaps’ is not yet found. There is a remarkable, probably traditional,102 extended usage of such absolute accusatives (or nominatives?) with a personal subject (Ach. 1182; Plut. 277?; Ar. fr. 664).

A typically comic, presumably colloquial, type of inner accusative is represented by Ach. 95 ναύφαρκτον βλέπειν ‘to look like a warship’, etc.103 The inner accusative of figurae etymologicae is used as a means of intensification (e.g. κράον κεκραγέναι, λῆρον ληρεῖν).104 Genitive105

The frequency of the absolute genitive in Aristophanes suggests that this was no purely literary construction (c.85 examples, more or less equally divided between early and late plays).

The genitive of comparison (c.120 examples) is hardly ever challenged by the alternative construction with ἢ (κρείττων ἢ ἀνήρ vs. κρείττων ἀνδρός). In some cases is of course necessary, for instance when an adverb, an adverbial phrase, a preposition, a full clause, or an infinitival construction follows;106 also πλεῖν is always accompanied by ἢ. Elsewhere the genitive of comparison is replaced by the construction with mainly for the sake of greater clarity.107 The usual preference for the genitive is a sign of syntactic conservatism.

vesp. 352 οὐκ ἔστιν ὀπῆς ‘there isn’t enough of a hole’ preserves an archaic syntagma with a partitive genitive instead of a nominative after a negative substantive verb.108 Apparently this construction had survived in colloquial language. Another colloquial feature is the frequent use of exclamatory genitives; again, this may be old.109

(p.253) Dative

The dative, which was lost in later Greek, is as strong in Aristophanes as elsewhere in classical Attic. The switch, with predicative participles, from the dative to the nominative (vesp. 133–5: ἔστιν δ’ ὄνοματῷνἱεῖβδελυκλέων, ἔχων τρόπονς φρυαγμοσεμνάκονς τινάς ‘the … son … is called … Bdelycleon: he’s got some rather haughtifalutin ways’; Av. 46–7) or to the accusative (Eq. 1394–5; Thesm. 674–6; Eccl. 1019–20: ταῖς πρεσβυτέραις γυναιξὶν ἔστω τὸν νέον ἔλκειν ἀνατεὶ λαβομένας τοῦ παττάλου ‘the older women shall be permitted, without penalty, to drag the young man away, taking hold of him by the peg’) may be a first symptom of weakening, but such phenomena are not uncommon in other writers either.110

4.1.2. Gender

Nominal gender is a topical issue in Clouds (see Ch. 4.2.2) and in the ‘women’s plays’ (see Ch. 6.6, especially on nominal motion). As in other languages with such a grammatical category, there are a few words with variable gender: τάριχος ‘dried fish’ mase. (Ar. fr. 207) vs. τάριχος ntr. (Eq. 1247), ὀρίγανος ‘origanum’ fern. (Eccl. 1030) vs. ὀρίγανον ntr. (Ran. 603a; Ar. fr. 128), λιμός ‘hunger’ fem. (Ach. 743: Megarian) vs. λιμός masc. (Pax 483).

4.1.3. Number

During the fourth century the dual completely disappears from Attic Greek, first in the verbal system, later also with nouns.111 Somewhat paradoxically, the dual is stronger in Euripidean tragedy than in Aeschylus, presumably because Euripides’ language is more open towards spoken Attic than the literary register of Aeschylus. It is clear from Aristophanic comedy that the dual was very much alive in fifth-century Attic. The difference between Aristophanes’ language and that of his younger contemporary Plato is striking: whereas Plato uses the dual in about 25% of all cases in which he could have used it, the corresponding Aristophanic figure is 57%.112 In the verbal paradigm the dual seems even stronger than in the nominal paradigm (c.75% observance). Many apparent exceptions are easily explained by the context (e.g. Pax 1054; Av. 95: the Hoopoe cannot know the number of people who are looking for him; also Ach. 729–835 in (p.254) Megarian).113 A tendency towards the loss of the dual can only be seen in the gradual increase of incongruous agreement, for instance between a dual subject and a plural verb; there are particularly many instances in the late Plutus.114

Since the dual is far less common in oratory than in comedy and consciously avoided in a writer like Thucydides (whose rate of observance is about 5%), it seems to have borne the stigma of Attic parochialism.115Hence, Aristophanes’ syntax is both conservative and dialectally purist with regard to the dual.

The σχῆμα Ἀττικόν (ntr. pl. subject agreeing with singular verb) is firmly observed.116 occasional variation between singular and plural in other contexts reflects the freedom of lively spoken language.117

Poetic plurals118 and the pluralis maiestatis (Thesm. 183, 196) are exclusively parodie.

4.2. Adjectives

There is only a handful of examples of a singular neuter adjective with the article in the function of an abstract noun (e.g. Nub. 414 τὸ ταλαίπωρον ‘endurance’, anap.; Plut. 562 τὸ σφηκῶδες ‘waspishness’, anap.); only the common τὸ δίκαιον ‘justice’ occurs in iambic dialogue (Ach. 500; Av. 1435).119 The neuter participle is not used in this way at all.

The replacement of a possessive genitive by an adjectival construction is parodic (e.g. Ach. 433 τῶν Θυεστείων ῥακῶν ‘of the Thyestean rags’; Av. 939; Thesm. 919; Ran. 1142).

4.3. Article

Aristophanes often omits the article in parodic passages (paratragedy, lyric parody) and less frequently in non-parodic lyrics with a solemn tone (e.g. (p.255) hymns); there is nothing comparable with the constant tragic omission for the sake of ‘de-automatization’.120 With certain nouns like ἐκκλησίαν or ἀγορά, the omission of the article (e.g. εἰσ ἐκκλησίαν, but ἐν τἧκκλησίᾃ121) is paralleled in prose so that it is difficult to find instances where the article is dropped metri gratia only.122

With ὃδε and οὗτοσ, however, the rules are less strict. The type ὃδε/οὗτος ἀνήρ is certainly less common than in tragedy, but it does occur, especially with ὃδε, even in (apparently non-parodic) comic dialogue. In Acharnians and Birds I have counted the figures set out in Table A. 2 (which are approximate only since it is sometimes difficult to decide whether ὃδε/οὗτοσ stands in an attributive relationship to a neighbouring noun).

TABLE A. 2. Frequency of ὃδε and οὗτος in Acharnians and Birds

Type ὅδε ὁ ἀνήρ


ὃδε ἀνήρ

οὗτος ἀνήρ










a Ach. 336, 454 (paratragic, trim.), 985? (trim.), 1191/2 (paratragic).

b Ach. 130, 462–3, 960, 1049 (all trim.).

c Av. 921 (trim.), 936–7?, 1274 (trim.), 1313, 1366 (trim.), 1725.

Particular uses of the article, which without doubt belonged to spoken Attic, include the address to slaves (e.g. Ran. 40 ὁ παῖσ; Ran. 271; also to other subordinated persons: Ach. 155 οἱ Θρᾶκεσ\ Av. 665 ? πρόκνη) and the strengthening of τί; in τὸ τί; (Nub. 775; Pax 826; Ran. 7, 40).123

As a relative, anaphoric, or demonstrative pronoun, the article is not used in Attic (except in idioms such as ὁ μέν/ ὁ δέ, τὸν καὶ τόν, πρὸ τοῦ); accordingly, it is restricted in Aristophanes to passages in other dialects.124

(p.256) 4.4. Pronouns

The variation between the older possessive construction ὁ ἐμὸσ πατήρ and the younger ὁ πατήρ μου is a clear case of syntactic polymorphy in Aristophanes. The older variant is still predominant (186 : 1 10).125Further details are given in Ch. 6.8.1.

The non-reflexive possessive construction rarely replaces the reflexive ἐμαυτοῦ, ήμέτερος αὐτῶν, etc., and when it does, there may be metrical reasons (Eq. 565; Plut. 55).

Unlike tragedy and occasionally prose, Aristophanes has no example for the change from ἐμαυτόν, σεαvτόv to a generalized reflexive pronoun ἑαυτόν, which became more common in Koine Greek.126

It is doubtful if there is an Aristophanic example for the later use of εἷς as indefinite pronoun (= τις).127

4.5. Prepositions128

As in Attic prose, but not in tragedy, σύν/ξύν is not used alongside μετά as a comitative preposition (see above, App. § 2.2.3).129 When a person is accompanied and when simultaneity is emphasized, ἄμα + dat. ‘together with’ is an alternative (also sociative αὐτόσ: Eq. 7 αὐταῖσ διαβολαῖσ ‘together with his lying-tales’). The participial construction with ἔχων/λαβών is very frequent, too. The prepositional adverb ὁμοῦ is common in food and cooking contexts (Eccl. 404; Ar. frs. 22, 383, 581. 6, 701).

The prepositions and prepositional adverbs ἀμφί, ἀνά, δίχα, χωρίσ, ὑπαί as well as μετά + dat. are confined to stylistically marked contexts (parody, lyrics, dialect parts) or to specialized usages (Ran. 554 with distributive ἀv’ ἠμιωβολιαῖα ‘in half-obol portions’); other prepositional words like ἄγχι, ἄτερ, πέλασ are completely missing

For prepositional ‘until’ Aristophanes has only μέχρι. (rarely), never ἄχρι, ἔοτε and ἔως which become more common in post-classical Greek.

The agentive preposition is nearly always ὑπό. Nub. 1122 πρὸς ἠμῶν and Ach. 226 παρ’ ἐμοῦ (lyr.) are isolated, agentive ἐξ non-existent in Aristophanes.

(p.257) In addition to the directional prepositions εἰς, ὡς, ἐπί, and πρός, Aristophanes has almost half of all the classical Attic examples of εὐθύ + gen. ‘straight to’; possibly this prepositional adverb had a colloquial touch.130 Directional ὡσ, which is rare in tragedy, becomes increasingly frequent in the two fourth-century comedies (Eccl., Plut.\ 18 examples, against only 16 in the six extant comedies written down to 414 BC: Ach.y Eq., Nub., vesp., Pax, Av.).

Except in the case of πέρι, the postpositional use of prepositions with anastrophe is parodie or consciously high-flown (e.g. vesp. 1118; Lys. 1145–6).131 The same applies to the positioning of the preposition between an attributive adjective or pronoun and its noun (e.g. Nub. 1159 τοῖσδ’ ἐνὶ δώμασι ‘within these halls’). However, anastrophic μέτα, πάρι, and ἔνι can be used, even in comic dialogue, as adverbial equivalents of μέτεστι, etc.

4.6. Verbs

4.6.1. Diathesis

Like nominal gender, verbal gender (diathesis) is comically exploited (Eq. 115 ῥέγκεται, 1057 χέσαιτο); foreigners cannot handle it correctly (see Ch. 7.5.3; also Pax 291132). Apart from this, there is little irregularity or variation. The passive future replaces the middle future in Eq. 98 (contrast Lys. 910),133 but with verbs carrying a suffixal -η- the inverse is the case (Nub. 1379 τυπτήσομαι; vesp. 1491; Ran. 797). The active imperatives παῦε ‘stop!’ and ἒγειρε ‘wake up!’ (?) may be very old relics.134

4.6.2. Tense

on the ‘resultative’ perfect (and the periphrasis ἔχω + part, aor.) see Ch. 5.4, on periphrastic constructions with a participle + εἷναι both in the perfect and in the present see Ch. 5.9. (and above, App. § Here, the Aristophanic plays foreshadow fourth-century developments, though not more strongly than other fifth-century texts.135

The expression of a ‘futur proche’ with μέλλω + inf. is not more common in Aristophanes than in contemporary prose. However, prose writers (p.258) prefer the construction with a future infinitive, whereas in Aristophanes, as in later Greek, that with a present infinitive is much more frequent (except after the past tense ἒμελλον).136

4.6.3. Aspect

A remarkable aspectual distribution exists in the imperative, where the present imperative is slightly predominant in human conversation (though less clearly than in tragedy), but the aorist imperative is preferred in addressing gods: see Ch. 2.9 for more details. The (epic) imperative οῖσε ‘bring!’, which is apparently built on the future stem and probably a comically pathetic form rather than a colloquial archaism, is aspectually isolated.137

4.6.4. Mood Indicative

The use of ἄν with the indicative of a past tense to indicate a habitual or repeated action is much more widespread in Aristophanes than in other fifth-century authors. It is therefore presumably colloquial but not necessarily as recent a development as its visible history might suggest (see further Ch. 6.8.1).

It is disputed whether a modalized future indicative + ἄν for emphatic future statements should be recognized as genuine (cf. e.g. Nub. 1157; vesp. 942; Av. 1314).138 Emphatic prohibitions with οὐ μή + fut. ind.139 are so common that the indicative occasionally spreads even into the domain of οὐ μή + subj. in emphatic denials (Pax 1037; Ran. 508–9). The use of μή + ind., which is characteristic of oaths (Av. 195; Lys. 917–18; Eccl. 1000; Ar. fr. 395), must be distinguished from (ὅπως) μή + ind. after explicit or implicit expressions of fear (Ach. 343; Nub. 493?).

For moods in final clauses see below, App. §§ and 4.8.6. OPTATIVE

Given the retreat of the optative in post-classical Greek, Aristophanes’ observance of the oblique optative in final clauses after a secondary tense shows his syntactic conservatism. The oblique optative is here used 40 (p.259) times and neglected 14 times (74%). While this figure is similar to that of contemporary oratory, a complete reversal has taken place by the time of Menander (c.27%).140 Moreover, the irregular use of the subjunctive is always due to the agent’s continued interest in the accomplishment of the action, so that the past tense in the main clause does not adequately express the true temporal relationship.141 The strength of the oblique optative is also obvious in its occasional intrusion into final clauses without a preceding past tense, mainly when the context of the main clause suggests some reference to the past.142

very rarely the potential optative occurs without the particle ἄν, a phenomenon that is ‘excluded only from the most rigid and fastidious sorts of writing’.143 Subjunctive

Unlike the omission of ἄν with the potential optative, the absence of ἄν in conditional, relative, or temporal clauses with the subjunctive is usually due to a pompous or parodie context;144 in anapaests and with πρίν it is slightly more common.145 Imperative

The function of the imperative is often taken on by other constructions such as rhetorical questions,146 indirect formulations (e.g. σὸν ἒργον ἐστί + inf. ‘it is up to you to’, often used by the chorus;147 independent ὅπως + ind. fut.: cf. below, App. § 4.8.6), the potential optative,148 the imperatival infinitive,149 or constructions with the verbal adjective in -τέο- (cf. Ch. 5.8).

(p.260) Some colloquial usages of the imperative are particularly frequent in Aristophanes. For instance, a second-person imperative often ‘agrees’ with πᾶς/τις150 although τις is used even more frequently in the ‘correct’ way with a third-person imperative in an address by a master to his slaves (e.g. Ach. 570–1, 805, 1096; cf. Lys. 1050 with πᾶς). Another idiomatic usage is the rhetorical question οἷσΘ’ ὅ δρᾶσον; ‘do you know what you should do?’, which contains, at least in synchronic terms, the aorist imperative (e.g. Eq. 1158; Pax 1061).151

4.6.5. Nominal forms of the verb

Exclamatory infinitives are far more common in Aristophanic comedy than in other genres. Like the genitive of exclamation they certainly belonged to colloquial Attic (e.g. Av. 5–6 τὸ δ’ ἐμὲ κορώνῃ πειΘόμενον τὸν ἄΘλιον ὁδοῦ περιελΘεῖν στάδια πλεῖν ἣ χίλια ‘To think, wretched me, that I’ve gone around and about for more than a hundred miles of travelling, at the bidding of a crow!’).152 Most of the exclamatory infinitives are preceded by the article (exceptions: Nub. 819?; vesp. 835).

on ‘optatival’ infinitives see Ch. 2.9, on ‘imperatival’ or ‘jussive’ infinitives App. § above, on the articular infinitive Ch. 5.9. Both the articular and the absolute infinitive (e.g. ἐμοὶ δοκεῖν ‘it seems to me’) are more common in literary prose than in comedy. Aristophanes has about four examples of the absolute infinitive.153

on the verbal adjectives in -τέο- see Ch. 5.8.

4.7. Particles

According to Denniston, particle usage in Aristophanes and in the orators is more purely Attic than that in the tragedians, Plato, Xenophon, and sometimes Thucydides.154 For instance, ‘ancillary’ οὗν with other particles (except in the combinations άλλ’ οὗν, γοὗν, δ’ οὗν, and μὲν οὗν) or οὐδέ without a preceding negative phrase are rare among the ‘pure Attic’ authors.

Differences between Aristophanes and tragedy must be due to the diverging stylistic levels of the two genres. Thus, δήπου and τοίνυν are frequent in comedy (and prose) but rare in tragedy, whereas interrogative ἧ is (p.261) far more usual in tragedy than in Aristophanes.155 only the base Scythian uses ναίκι (= ναίχι) as an emphatic assentient particle (cf. Ch. 7.5.4).

In certain passages Aristophanes mocks a current fashion of particle usage (Eccl. 773–6 assentient γάρ; Plut. 833–8 κομιδἧ μὲν οὗν‘absolutely’; on Ran. 559–67 see Ch. 6.8.1) or uses a particular particle for a stylistic end (αὐτΆρ in a hexametric oracle: Av. 983; μέν at the beginning of a speech: Ach. 136; vesp. 907; Thesm. 383; Plut. 489).

The widespread view that a high frequency of particles was typical of spoken and more colloquial Attic has recently been modified by Duhoux.156 Whereas in fifth-century drama (both tragedy and comedy) particles constitute only about 13% of all words (Menander c.10%), the corresponding figure for Plato’s and Xenophon’s dialogues is c. 17–18%. Within Aristophanes there is no general statistical difference between dialogic and non-dialogic passages, but γε or δῆτα are especially strong in dialogue, δέ, ἥ, καί, οὐδέ, and τε in other parts.

4.8. Sentence Structure and Complex Sentences

4.8.1. Nominal sentences

Nominal sentences are more frequent in Aristophanes (Lys.: 60.0% nominal sentences: 40.0% copular sentences) than in Plato (Symp.: 53.6% nominal sentences), Lysias (1, 7, 10, Eroticus: 58.7%), Demosthenes (De corona: 46.7%), or Isocrates (Paneg.: 50.6%), but less frequent than in tragedy (Soph. oT: 72.3%) and Thucydides (bk. 2: 77.8%).157 obviously nominal sentences were perfectly common in spoken Attic and not confined to particular genres or authors.158 It is impossible to state confidently how ‘conservative’ or ‘innovative’ Aristophanes was: his percentage seems high in comparison with fourth-century prose, but this may be explained by the difference of text genre.

4.8.2. Parataxis

The high incidence of parataxis in comedy certainly reflects the general preference of spoken language for paratactic (and sometimes asyndetic) sentence structures. Even in comedy, however, parataxis with καί, δέ, and other simple connective particles is used in particular to produce simple and unpretentious narratives.159 From such passages, connecting (p.262) κᾆτα/κἄπειτα was transferred into the position between a conjunct participle and the main verb, a construction that is quite frequent in comedy (e.g. Eq. 391–2 τοιοῦτος ὤνκᾆτ’ ἀνὴρ ἔδοξεν εἷναι ‘having been that sort of character, he’s now managed to pass for a man’).160

4.8.3. Substantive clauses

Declarative verbs like λέγω can be followed either by a substantive clause introduced by ὅτι or ὡς or by an infinitival construction (λέγω τὸν ἄγγελον ἥκειν). The former variant is strongly assertive, whereas the speaker’s attitude and judgement are left more open with the infinitive.161 Apart from the slight semantic difference, the choice is also determined by generic considerations. The percentages, given in Table A.3, for substantive clauses after λέγω are based on Fournier’s statistics.162 In general the infinitival construction seems more elevated and appropriate for literary prose. Rather surprisingly, however, it is also considerably more frequent in Aristophanes than in the orators. This may indicate that Aristophanes’ language is not entirely ‘oral’ (but a more thorough investigation taking into account other factors such as sentence length would be needed to substantiate this conclusion).

TABLE A.3. Substantive clauses after λέγω in classical Greek (%)

































A certain artificiality of Aristophanic language is further indicated by the choice between ὅτι and ὡς (ὅπως only in Pax 132, paratragic), which are (p.263) semantically interchangeable (Table A.4).163 Apparently, ὡς was an Ionic (Herodotus) and poetic (tragedy) form, which was not very common in unpretentious Attic, as the figures for Lysias suggest. The relative frequency of ὡς in Aristophanes must then be explained by the fact that such a metrical alternative was extremely convenient.

TABLE A.4. Frequency of ὅτι and ὡς introducing substantive clauses in classical Greek





















4.8.4. Indirect questions

In classical Attic indirect ‘who?’-questions are introduced most commonly by ὅστις, more rarely by ὅς. Later, the direct interrogative pronoun τίς becomes predominant. The distribution in different authors is shown in Table A.5.164 The fact that τίς is quite common in Lysias but absent from Aristophanes illustrates again Aristophanes’ conservatism. With indirect (p.264) questions other than ὅςτισ/τίς the direct interrogative pronoun is found in Aristophanic comedy as well.165

TABLE A. 5. Frequency of indirect questions in classical Greek


























4.8.5. Relative clauses166

In relative clauses with defining value (i.e. where the subordinate clause defines a conceptual class to which the antecedent belongs), both ὅστις and ὅς can be used, but again the preference for ὅστις is much stronger in comedy than in other genres, including the speeches of Lysias.167 Since ὅς is particularly frequent in Herodotus, it may have been prominent in Ionic and spread from there into Attic. Aristophanes would then keep his Attic pure again.

In addition, the preference for bisyllabic pronouns (cf. above, App. § 3–3.3) supported ὅστις. Accordingly, οἷος/ὅσος are more frequent than ὁποῖος/ὁπόσος (but among the relative adverbs, the short forms οὗ/ᾗ/οἷ predominate over ὅπῃ/ ὅπου/ὅποι).168 of course, the semantic context also plays a role in all these cases, and the differences between the various genres are insignificant.

4.8.6. Final clauses

The frequency of ὅπως (ἄv), ὡς (ἄv), and ἵvα in introducing final clauses has been discussed in Ch. 6.8.1. The figures shown in Table A.6 have been counted by various scholars.169 In the present context, the most important observation is that Aristophanes, unlike Lysias, still uses the traditional Attic ὅπως (ἄv) beside ἵvα. Even more surprisingly, ὡς ἄv is occasionally found in non-parodic trimeters (e.g. Ach. 44; Eccl. 57; Plut. 112); without doubt, this is at least partly due to metrical convenience.

In negative final clauses, ἵvα μή (47 exx.) easily surpasses μή alone (7), ὅπως (ἄν) μή (6), and ὡς (ἄv) μή (Av. 1509). In iambic trimeters there are only 3 cases of μή and ὅπως μή respectively.170 However, simple μή is normal (p.265) with verbs of fearing and verbs of caution/intention/effort (which have ὅπως as their positive conjunction): 29 examples of μή stand against 5 of ὅπως μή (Eq. 112, after a verb of fearing; vesp. 141, 155, 372; Thesm. 653) and 1 of ὡς μή (vesp. 113).

TABLE A.6. Frequency of final conjunctions in classical Greek


ὅπως ἄν


ὡς ἄν














































Ten orators





With verbs of intention and effort (φνλάττω, φροντίζω, μεμνῆσθαι) the use of ὅπως + fut. ind. is regular. From here, the future indicative spreads occasionally—but in comedy more frequently than in prose and tragedy—into final clauses in a strict sense.171 Conversely, a subjunctive after ὅπως ἄν occurs after verbs of intention and effort.172 Such grammatical flexibility is characteristic of spoken language.

Imperatival independent subordinate clauses introduced by ὅπως (e.g. Eccl. 149 ὅπως ἐρεῖς ‘make sure that you speak’) are extremely common in Aristophanes, who has c.40 out of the 80 examples in classical Greek. obviously this was a colloquialism.173

4.8.7. Consecutive clauses

Consecutive ὡς (for ὤςτε), which is common in tragedy, is found only once in Aristophanes (Ran. 1110). The usual ὤστε is construed with either an infinitive or a finite verb (mostly the indicative either as realis or as irrealis with ἄν, but also the potential optative). The infinitive construction presents the consequence in a more abstract manner than the vivid and (p.266) concrete personal mode,174 but the semantic difference does not fully explain the wide divergence between tragedy and literary prose (Herodotus) on the one hand and Aristophanes and oratory (Lysias) on the other. The variant with the personal mode must have been predominant in spoken Attic (Table A.7).175

TABLE A.7. The construction οf ὤστε in classical Greek

% ὤςτε (ὡς)

+ inf.

+ pers. mode






62.1 (!)













4.8.8. Conditional clauses

on the variation between ἐάν, ἥν, and ἄν see App. § 2.1.5 above, on paratactic conditional clauses see Ch. 6.8.1. The various (regular) hypotactic types in Aristophanes have been listed and discussed by Sobolewski.176

4.8.9. Causal clauses

Statistics for the use of the causal conjunctions ὅτι, ὡς, and διότι have been compiled by Monteil (Table A.8).177 The easiest way to interpret these figures and to relate them to the use of ὅτι/ὡς as substantivizing conjunctions (cf. above, App. § 4.8.3) is to assume that Attic originally used only ὡς in causal clauses, but in substantivizing clauses both ὅτι and ὡς, whereas Ionic preferred substantivizing ὡς and causal ὅτι The two conjunctions became completely interchangeable with the growing influence of Ionic on Attic (and vice versa) although, to judge from Lysias, ὅτι was generally preferred in simple and non-literary Attic (in accordance with the tendency towards bisyllabic conjunctions). If this is correct, both the high number of causal ὡς and the avoidance of the recent variant διτι in Aristophanes again (p.267) underline his conservatism, especially when his usage is compared with that in the unpretentious oratorical prose of his contemporary Lysias.

TABLE A.8. Frequency of causal conjunctions in classical Greek
























Aristophanes also has c.50 examples of causal ἐπεί and c.30 examples of causal ἐπειδή.

4.8.10 Temporal clauses

The most common temporal conjunctions in Aristophanes are ὅτε/ὅταν (48/83 exx.), ἡνίκα (ἄν) (25/17), ἐπειδή/ἐπειδάν (9/15), ὁπότε/ὁπόταν (2/12), ἐπεί/ἐπήν (4/3), ἕως (ἄν) (6/18), ὡς (17)> ἐξ οὖ/ἐξ ὅτου/Άά’ οὖ, etc. (19), and πρίν (ἄν) (26/58, of which 57 are with the infinitive). In the mock oracle of Av. 983 and in the ‘law’ of Av. 1355 ἐπήν is an archaism; in Lys. 1175 it must give a solemn note to Lysistrata’s words. The quasi-Hellenistic temporal conjunction ἐξ ὅτε at Av. 334b is unique and emendation may be necessary.178

4.9. Conclusion: Syntax

To paint a coherent general picture for Aristophanes’ syntax is more difficult than for his phonology and morphology. Some syntactic features are introduced for parodie purposes and can be discarded in our context (poetic plurals, omission of the definite article). Among the rest there are no features whose existence in spoken Attic is entirely implausible, but it is likely that the frequency of λέγω with the infinitive or of ὡς as a substantivizing conjunction were furthered by metrical requirements (similarly ὡς ἄν in final clauses, omission of ἄν with the subjunctive and potential optative).

There is a series of characteristics that are more common in comedy than in other literary genres. These illustrate the colloquial layer in Aristophanes’ language (e.g. frequency of exclamatory genitives and (p.268) exclamatory infinitives, frequency of paratactic structures, predominance of ὥςτε + personal mode, imperatival independent ὅπως-clauses, πᾶς/τις + second-person imperative, past indicative + ἄν for habitual actions, εὐθύ + gen. ‘(straight) to’, inner accusatives with βλέπω). None of these features, however, is necessarily recent.

The innovative features in Aristophanes’ syntax (growing frequency of transitive active perfects, μέλλω + pres. inf.) are easily outnumbered by the conservative features in other areas (frequency of dual, strength of oblique optative, genitive of comparison rather than ἢ, retention of final ὅπως ἄν, frequency of causal ὡς and absence of causal διότι, absence of τίς in indirect questions). In the use of the dual, syntactic conservatism goes hand in hand with a certain Attic purism (cf. also particle usage, preference for ὅστις in ‘defining’ relative clauses).

5. General Conclusion

The preceding tour d’horizon of Aristophanic grammar has shown that the same general tendencies prevail in all three classical areas of grammar: phonology, morphology, and syntax. Given that Aristophanes was regarded as the comic author par excellence by the ancient grammarians and Atticists, it is not surprising that linguistic conservatism and dialectal purism are the two most outstanding features of his language.

It is important to keep this in mind when searching, for instance, for the first symptoms of Koine-ized Greek. While it is true, as López Eire has shown,179 that some early evidence for later linguistic phenomena comes from the comedies of Aristophanes, these plays are also our first extensive source of colloquial Attic, a linguistic stratum which must have contributed its share to the formation of Koine Greek. Since colloquial features are not necessarily innovative features, it would be wrong to conclude that Aristophanes’ Attic is in any sense ‘later’ or more ‘diluted’ than that of contemporary writers such as Euripides, Thucydides, or Lysias. In fact, most of our evidence, in the form of both grammatical material and ancient opinion, suggests the exact opposite.

Moreover, there is an anthropological point to this. More than any other literary genre in fifth-century Athens, comedy reflected, and contributed to construct, the local citizens’ corporate identity. In this function old Comedy parallels carnivalesque genres in other cultures, where in-group language often enforces social cohesion.180 Aristophanes wrote as an (p.269) Athenian, for Athenians, and on Athenian matters. Could there have been a more suitable medium than the purest Athenian language? (p.270)


(1) For this terminology see Björck (1950) 84–91.

(2) Cf. e.g. Ar. test. 65 (Quint. 10. 1. 65–6), test. 69 (Phot. Bibl. 158, p. 101b4 Bekker), test. 87 (Schol. Thuc. 1. 30. 1).

(3) These aspects are treated extensively by López Eire (1996a), (1997), and (1998); cf. already Lottich (1881). on Aristophanic imagery see Taillardat (1965); on obscenity Henderson (1991), Dover (2002), and the literature cited by Willi (2002a) 10–11; on comic word formation Uckermann (1879), the articles by Peppier (1910), (1916), and in particular (1918), (1921), as well as, for the compounds, G. Meyer (1923) 119–32, and Costa Ramalho (1951).

(4) For even more eclectic (but good) surveys see Anagnostopoulos (1925), Hoffmann-Debrunner-Scherer (1969) 116–26, Hiersche (1970) 163–77, and López Eire (1986). Most of the material presented here was first collected in Willi (2001) 168–97.

(5) Also in ὦ τᾶν ‘my friend’, which may originally be, with Kretschmer (1909) 58, an allegro variant of ὦ τάλαν; but cf. also Dickey (1996) 159–60.

(6) Cf. for (2) Lautensach (1911) 200–2; for (3) Chantraine (1933) 31–2 and 380–2, Petersen (1937), and Björck (1950) 260–74; for (4) Björck (1950), esp. 46–56; and for (5) Colvin (1999) 137–40.

(7) Wackernagel (1887) 119. Pace Monteil (1963) 267, οὕνεκα is never used as a causal conjunction in comedy: in Nub. 422 (anap.) οὕνεκα or εἵνεκα goes with τούτων.

(8) Wackernagel (1887) 109–15, criticized already by Coulon (1908) 26–32.

(9) Cf. B. Rosenkranz (1930) 149. Similarly, it may be over-zealous to replace, with Sobolewski (1890) 99–100 and Coulon (1908) 32–5, all the variants in -εν: see Thumb-Scherer (1959) 296.

(10) e.g. Pax 140; Thesm. 1122; cf. Rutherford (1881) 432 n. 1; Sobolewski (1890) 34–63; Coulon (1908) 91; C. Austin (1973) 133 on the possible exceptions at vesp.147 and Thesm. 657; Colvin (1999) 207.

(11) But note the common ὄρνεoν ‘bird’; Av. 368 oνγγενέε (trim.) is suspect and Ach.106 Ἰάоνας imitates the foreigner talk of Ach. 104.

(12) e.g. Ach. 1185, 1191/2; Pax 1064; Av. 972, 978; Thesm. 108, 111, 126; Ran. 1353·

(13) e.g. Eq. 973; Nub. 297, 401?; vesp. 615; Av. 1748; Ran. 213, 675.

(14) e.g. Pax 1098; Av. 246/7, 698; Thesm. 325, 1044.

(15) Cf. Threatte (1980) 271–4.

(16) Cf. Anagnostopoulos (1925) 16; Lupas (1972) 42.

(17) Cf. Kretschmer (1888) 479; Threatte (1996) 183.

(18) Buck (1955) 106; correctly B. Rosenkranz (1930) 147–8; Thumb-Scherer (1959) 296–7; Hoffmann-Debrunner-Scherer (1969) 111.

(19) Threatte (1996) 672–4.

(20) The manuscripts have ἂν in Av. 554; Lys. 901, 902; Thesm. 154, 1187: Sobolewski (1891) 13; cf. Arnott (2002) 193–4.

(21) Threatte (1996) 315–17.

(22) According to the statistics in Lupas (1972) 42.

(23) Unlike the inscriptions and Thucydides (cf. Schwyzer (1939) 607; Threatte (1996) 314) Aristophanes never uses the older σφῶν αὐτῶν instead of ἑαυτῶν, nor the indirect reflexive pronouns οὖ/ού, οἷ/οἱ, etc. on their own, except in Nub. 1313 (lyr.); cf. Kallenberg (1925) 64–79.

(24) Lupaş (1972) 48–9·

(25) Cf. Lupaş (1972) 49–52, also on οἷος, etc.; Ussher (1973) 170–1 on στοιά ‘hall’, ροιά ‘pomegranate’, ποία ‘grass’; Dover (1993) 209 on ἠλόησεν ‘threshed, struck’; for later comedy Arnott (2001a).

(26) Threatte (1980) 326–9.

(27) Buck (1955) 69–70.

(28) Cf. Allen (1987) 61 and Colvin (1999) 266, against (e.g.) Platnauer (1964) 71.

(29) e.g. Nub. 567/8; Thesm. 988, 999; Ran. 328, 340/1.

(30) See Colvin (1999) 166–7.

(31) e.g. Pax 1286; Av. 238, 250.

(32) e.g. Ran. 992, 1172, 1314; in monodies: Thesm. 1052; Ran. 1349; occasionally elsewhere with reference to tragedy: Ran. 827, 898.

(33) Sobolewski (1890) 32–4: σύν is demanded by the metre 6 times, ξύν 2 times, συν- 86 times, ξυν- 23 times; cf. Colvin (1999) 209–10; for the inscriptions Threatte (1980) 553–4.

(34) e.g. Nub. 604; vesp. 1081; Thesm. 102, 1034; Ran. 445. Lys. 1143 is exceptional.

(35) e.g. Nub. 335, 401; vesp. 570, 678; Av. 579, 591, 686; Ran. 680; more examples in White (1912) 365, and L. P. E. Parker (1997) 92–3.

(36) Figures based on Allen (1987) 109, who refers to Schade (1908).

(37) Thus Eq. 207?; Nub. 869?; vesp. 151?, 837; Pax 1201; Av. 45, 820?; Plut. 166, 1019, 1153; cf. White (1912) 364–5.

(38) Widespread crasis is suggested not only by metrical considerations and the manuscript transmission but also by the fossilization of a crasis form in ὁ θάτερος(Men. fr. 491; cf. θάτερον = τὸ ἔτερoν).

(39) Krüger (1879) 23–30 gives an extensive list of examples; cf. also López Eire (19960)79-84.

(40) Cf. Platnauer (1960) 140.

(41) Menander has more than 30 cases of οὐδὲ εἷς; cf. in general Wackernagel (1928) 114, 268–9; Moorhouse (1962) 245–6. Separated οὐδ’ ἂν εἷς occurs already in Eq. 573 and vesp. 72. See also Stinton (1977) on interlinear hiatus, where Aristophanic comedy is much freer than Euripidean and Sophoclean (but not Aeschylean) tragedy.

(42) Cf. Zacher (1899) 465–73; Thumb-Scherer (1959) 293; Threatte (1980) 640–3.

(43) Cf. Threatte (1980) 377–83 and (1996) 213; the middle ending is absent from the relevant inscriptions (see Threatte (1996) 451–2), but Zacher (1899) 473–84 argues for -ει in Aristophanes; similarly Arnott (2001b) 40 for later comedy.

(44) See Threatte (1996) 239–42.

(45) Hiersche (1970) 165–6 makes a good case for σπ-/σκ-.

(46) Perpillou (1984): because of the sexual meaning of παίειν ‘to bang’, Пαιονίδης (Lys. 852) need not evoke πέоς ‘penis’, and in Eq. 1284–6 there is no link at all between αίσχραῖς ‘obscene’ and (ἐσχάρας ‘hearths’. See also Ch. 7.3.3 on the pronun ciation of ει.

(47) e.g. Ach. 276–83 Фαλῆς/(κραι)πάλης/βάλλε/βαλεῖς; Ach. 751 διαπεινᾶμ-/διαπίνομ-;cf. the warning by Merry (1893) about the use of Pax 926 βοΐ/βοη(θεῖν) as an argu ment for early iotacism.

(48) Cf. Dover (1981) 4, 14 n. 10; López Eire (1993) 49; Threatte (1996) 25–32.

(49) For the handful of exceptions see Gomme-Sandbach (1973) 735.

(50) only the locatival use (Ἀθήνησι remains current: see Threatte (1996) 96–101, 374–5.

(51) Nub. 306 (alongside ‘Doric’ a); Lys. 775 (‘para-epic’ oracle); Thesm. 39 (Agathon’s servant), 1148 (hymnic). Eq. 163 λαῶν (not guaranteed by the metre) may allude to Il. 4. 90: Neil (1901) 29.

(52) Cf. La Roche (1893) 222–6.

(53) e.g. Pax 1301 τοκῆας ‘parents’ (para-epic); Av. 679, 1412 (lyric voc. ἀηδοῖ ‘nightingale’, χελιδοί ‘swallow’); Lys. 774 πτερύγεσσιν ‘wings’ (mock oracle).

(54) e.g. Ach. 93 πρέσβεως ‘ambassador’ (not πρεσβευτοῦ); Nub. 559 εἰκούς ‘similes’ (ace. pl., after the s-stem comparative?); Nub. 1327 πατραλοϊα ‘father-beater’ (voc); Ar. fr. 26 φρυνώνδα (voc); Ar. fr. 318 ἥρως (nom. pi.), fr. 712 ἥρων (ace sg.).

(55) See Threatte (1996) 311–12; cf. Schwyzer (1939) 536 n. 3.

(56) ἀμείνων, βελτίων, ἐλάττων (Aristophanes never uses ὀλείζων like the inscrip tions: cf. Dover (1981) 4), ἡδίων, ἥττων, καλλίων, κρείττων, μείζων, ττλείων, χείρων.

(57) Cf. Peppier (1918) 183; Leumann (1945) 10–14.

(58) Cf. further Eq. 1252; vesp. 550; Pax 402; Thesm. 816–17. on Eccl. 1131 see Ch. 6.7.

(59) The latter might parody metrical lengthening in a hexameter. For irregular comparatives and superlatives in other comic poets see Peppier (1918) 181–3.

(60) In Ran. 1399 both τοιούτο and τοιούτον are transmitted, and in Pax 759 and Lys. 485 the transmitted form in -o is metrically impossible.

(61) Cuny (1906) 14; Threatte (1996) 91–3.

(62) Cf. Cooper (1972).

(63) Threatte (1996) 340–2; the Aristophanic figures are based on the index compiled by o. J. Todd (1932) and therefore refer to the edition by Hall-Geldart (1906) and (1907).

(64) Martín de Lucas (1996) 161–3.

(65) Dover (1997) 64; cf. Threatte (1996) 411–12.

(66) Cf. Denniston (1950) p. lxxvi, 287; τιή; is further strengthened as τιὴ τί δή;: vesp. 1155; Pax 1018; Thesm. 84.

(67) For the figures see Lautensach (1896) 26–8 (who counted 9 -μεςθα : 62 -μεθα in the then accessible fragments of Middle and New Comedy).

(68) See further La Roche (1893) 138–48; Lautensach (1916) 104–10, with statis tics.

(69) Ran. 1448 σωθείημεν ‘we would get saved’, if correct, probably quotes Euripides (fr. 582 Nauck).

(70) Lautensach (1917) 169–80 (but note Cratinus fr. 183 ἀμύναιν?); cf. La Roche (1893) 132–7; Duhoux (2000) 224–5.

(71) Plut. 505 (anap.) is doubtful and in Ar. fr. 332. 15 λέξαις is now retained.

(72) -ατo: Eq. 662; Nub. 1199; Pax 209; Aυ. 1147; Lys. 42; cf. Lautensach (1916) 113–16; Wackernagel (1916) 89–93.

(73) e.g. Pax 197; Eccl. 238, 771–2, 838; also Eccl. 265 εἰθισμέναι ἐσμέν ‘we are used to’. Note Ran. 1113 ἐστρατευμένοι εἰσίν ‘they are old campaigners’ vs. Ran. 1116 παρηκόνηνται, ‘they are sharpened’; similarly vesp. 1114 εἰςἰv ἐγκαθήμενοι vs. Ach. 343 ἐγκάθηνται ‘they sit inside’; cf. La Roche (1893) 171–2; Lautensach (1896) 29; Kontos (1898) 270–9, with a list for the classical Attic writers. In the active perfect subjunctive and optative, non-periphrastic forms like πεποιθοίη and κεκλόφωσι are usual: cf. La Roche (1893) 161–9; Lautensach (1917) 181–4. For the periphrastic medio-passive perfect subjunctive and optative see e.g. Ach. 671; Lys. 567; Thesm. 397; Plut. 680.

(74) Cf. Rutherford (1881) 242–3; Lautensach (1896) 4–5; E. Harrison (1942); Barrett (1964) 292–3.

(75) Cf. La Roche (1893) 220–1; Lautensach (1896) 6–11; Dover (1968) 144.

(76) See Lautensach (1918) 83–90, with statistics (Aeschylus 2:3; Sophocles 7 : 11; old Comedy (except Aristophanes) 1 : 7; Middle-New Comedy 2 : 6; but Euripides 10 : 3; Aristophanes 13:7).

(77) e.g. Pax 1276–92 passim; Aυ. 701, 777; Ran. 217; cf. Lautensach (1899) 165–81. Ach. 754 ἐμπορευόμαν ‘I set out’ should probably be corrected.

(78) Cf. Lautensach (1899) 2–3; Duhoux (2000) 96–7; Arnott (2002) 196–8.

(79) In Aristophanic comedy θέλω is only used in the idiomatic ἢν θεòζ θέλῃ ‘god willing’, etc.

(80) Cf. Lautensach (1899) 131–5.

(81) Cf. La Roche (1893) 155–60; for the 5th-cent. prose writers B. Rosenkranz (1930) 152. Lys. 895 διατιθεῖς (MSS) must stand for διατίθης.

(82) Cf. Lautensach (1911) 4–5 and (1917) 190–1.

(83) Threatte (1996) 600–2, 615–17; Lautensach (1911) 118–19.

(84) Contrast vesp. 574, 717; Lys. 108; Ran. 1190.

(85) Contrast e.g. Ach. 1022; vesp. 1107; Eccl. 116, 395.

(86) Cf. Lautensach (1911) 101–15.

(87) Lautensach (1911) 165–6, with statistics.

(88) Compare e.g. Nub. 185, Eccl. 1161, and Aυ. 697 with vesp. 1142, 1413, and Ar. fr. 663; similarly δεδοικώς (Pax 607; also Eccl. 181 δεδοίκατε ) vs. δεδιώς (Eccl.643; Plut. 448). For more detail, also on the following points, see Lautensach (1921) 231–48.

(89) Cf. Schwyzer (1939) 785; Hauri (1975), esp. 171–2.

(90) Excessive regularization, as promoted by Walker (1894) and (1906), is dangerous. The model πίπτω/ πεσοῦμαι/ἔπεσον must account for the aorist ἔχεσον (Thesm.570); cf. Schwyzer (1939) 786.

(91) Nub. 562; Ran. 737; cf. vesp. 726 δεδόκησαι; Ran. 1485 δοκήσας.

(92) Contrast e.g. Ran. 193 with vesp. 138; Humbert (1960) 153 sees here a ‘futur déterminé’ and a ‘futur indéterminé’.

(93) In trimeters: Nub. 792; vesp. 784; Aυ. 1456, 1506; Plut. 65; elsewhere: Ach.295; Nub. 1440; vesp. 437; Pax 1274; Aυ. 346; Lys. 262–3, 1280; Ran. 1047, 1106?; cf. Wackernagel (1928) 172–4, also on the origin of ὑπό τι ‘a little’ (vesp. 1290); Morpurgo Davies (1985) 86–8; Dunbar (1995) 697.

(94) e.g. Nub. 182, 1465, 1477; cf. Threatte (1996) 173–6.

(95) Cf. Dunbar (1995) 152, against Lautensach (1911) 10–11.

(96) Lautensach (1899) 47–9, 146–9, rejected by Mastronarde (1989).

(97) Pace La Roche (1893) 215–17: vesp. 900; Thesm. 649, 878; Ran. 921; cf. exclamations like κακοδαίμων ἐγώ ‘poor me!’: Dittmar (1933) 22–4.

(98) Cf. Dickey (1996) 199–206: Aristophanes has with 80% of his vocatives (cf. orators c.90%, Thuc. 85%, but later Men. 12%), but the view of Scott (1905) 39–41 that omission of in Aristophanes regularly indicates elevation of style is unfounded (cf. also Ch. 2.5).

(99) Aυ. 952; Thesm. 1045–6; non-parodic: Nub. 299/300 (lyr.); quotation: Nub.30; cf. Bers(1984)73.

(100) Perhaps Aυ. 941/2 ἀλᾶται ςτρατῶν ‘he is an outcast from the host’ (?): cf. Bers (1984) 100; Dunbar (1995) 537·

(101) Nub. 272; Lys. 1298; Thesm. no, 126, also 1055?; quotations: Ran. 1318, 1403; note locatival μεγαροῖ (Ach. 758) and Πυθοῖ (Lys. 1131 ≠Eq. 1273 Πυθῶνι) but also ἐv μαραθῶνι (vesp. 711; Ar. fr. 429); cf. Bers (1984) 95–8.

(102) Cf. M. L. Rosenkranz (1964) 271–3.

(103) Cf. Ach. 254; Eq. 631, 855; vesp. 455, 643, 847, 900; Pax 1184; Aυ. 1169, 1671; Ran. 562, 603a, 804; Eccl. 292b; Plut. 424.

(104) e.g. Eq. 487; Aυ. 31, 1401; Thesm. 793, 880; Eccl. 106; Plut. 419, 517, 1044; cf. vetta(1989) 153.

(105) See the comprehensive study by Poultney (1936), with lists for all the usages.

(106) e.g. Nub. 1418; vesp. 650; Pax 23; Lys. 304; Thesm. 519; Ran. 947; Plut. 939; note Eccl. 700–1 πpότερον αὐτῆς for πρότερον ἢ παρ’ αὐτῇ). A substantivized infinitive in the genitive of comparison occurs only once: Aυ. 1343a. For general statistics see Schwab (1893-4) ii·42-3, 78, and 92 (where the ratio of 9.8 gen. : 1 in comedy would be even more in favour of the genitive if πλεῖν ἢ had been excluded; cf. e.g. Aesch. 35.5 : 1, Soph. 9.4 : 1, Eur. 11.5 : 1, Hdt. 19.8 : 1, Thuc. 11.6 : 1, Xen. 3.8 : 1, Plato 7.6 : 1, Dem. 3.9 : 1).

(107) Nub. 432; vesp. 1068/9, 1149; Aυ. 369; Lys. 432; Thesm. 273; Eccl. 1040, 1096. A possible exception is Thesm. 203 with ἢ σύ instead of σοῦ.

(108) Cf. Schwyzer (1940) 10.

(109) Examples in Dittmar (1933) 24–5; Poultney (1936) 125–6.

(110) These sentences are not anacolutha like Ach. 1164–8 or Pax 1242–4: for the latter cf. Slings (1992) 96–101. The complete loss of the dative occurs much later: see Humbert (1930).

(111) Cf. in general Hasse (1891) and Cuny (1906); for the Attic inscriptions Meisterhans-Schwyzer (1900) 199–201.

(112) Cuny (1906) 94, 163–4; cf. Bers (1984) 59–60; Duhoux (2000) 131–4. López Eire (1991) 21–6 exaggerates the weakness of the Aristophanic dual because he does not compare it with other writings.

(113) But metrical need can be a determining factor, too: e.g. Aυ. 367 παθόντες/ἄνδρε; Ran. 605–6 ξυνδεῖτε/ἀνύετον.

(114) Plut. 73, 429–30, 471, 482, 484, 509, 532, 608–9, 733–4, 735–6; cf. Poultney (1963)363-7.

(115) Meillet (1965) 218.

(116) Cf. Kühner-Gerth (1898/1904) ii/1. 64–6 (also on ‘regular’ exceptions with personal subjects, as in Ach. 806–7); Wackernagel (1926) 101–3; Poultney (1963) 362–3. In Pax 1262 the plural verb is due to the separation from its subject.

(117) e.g. vesp. 433, 975–6; Lys. 209, 763; Eccl. 504–10; cf. the constructio ad sen- sum in Eccl. 431–2. on singular vs. plural verbs used by/to the chorus see Kaimio (1970) 36–157: in the first person Aristophanes makes the chorus use the plural more often than the tragic writers.

(118) e.g. Ach. 426, 479; Nub. 1165; Thesm. 41, 874; Ran. 1339a; Eccl. 11.

(119) Elsewhere e.g. Eccl. 895 τὸ σοφóv ‘expertise’ (cf. Lys. 547); Eccl. 901 τὸ τρυφερόν ‘voluptuousness’.

(120) e.g. Nub. 275–90, 298–313 (but not Nub. 949–58, 1024–33); Aυ. 904–53 pas sim, 1372–1401 passim; Thesm. 101–29, 695 (paratragic); Eccl. 507–8 (paratragic); cf. also Ch. 2.10. on tragic ‘de-automatization’ or ‘defamiliarization’ see Bers (1984) 190–4.

(121) Cf. Ussher (1973) 94, 108, 112; in general Cooper (1998) i. 390–1.

(122) Anagnostopoulos (1925) 16 adduces Lys. 18, 101, 882, where the omission is in fact surprising, but all of these passages would mean something slightly different if there were an article; note however Lys. 108 μιλήσιοι.

(123) Cf. Wendel (1929) 8; Svennung (1958) 214–24;τὸ τί;: Humbert (1960) 51–2. The type ὁ παῖς is not ‘solemn’ as Del Corno (1985) 158 claims.

(124) Cf. Witkowski (1915) 24–5; Colvin (1999) 226–7; also Eq. 1039 and Aυ. 973 (mock oracles); Thesm. 126 (Agathon’s song).

(125) A certain reluctance towards the younger construction may account for the irregular replacement of μου by ἐμοῦ in Lys. 301 and for the intrusion of the enclitic personal pronoun into the position of the possessive adjective (Aυ. 1110; Lys. 168, 416). Conversely, the reflexive possessive genitive αὑτοῦ/αὑτῆς can stand in enclitic (or proclitic) position: Nub. 515–16, 905; Pax 880; Aυ. 475.

(126) Cf. Schwyzer-Debrunner (1950) 197–8; Lys. 1070 comes closest.

(127) Cf. Eq. 1128; Aυ. 462, 1292; Wackernagel (1928) 318.

(128) For more detail see Sobolewski (1890).

(129) Cf. Mommsen (1895) 1–9, 634–60.

(130) Richards (1901): 25 cases in classical Attic (10 in Aristophanes, 5 in other comic writers, 9 in Plato, Thucydides, and Xenophon taken together).

(131) Cf. Mommsen (1895) 644–5, 653–4; Wackernagel (1928) 196–7; Bers (1984) 17 on Arist. Poet. 1458b 31–4.

(132) Cf. Bers (1984) 111–12; Cassio (1985) 149; López Eire (1986) 243; Colvin (1999) 291–2; Duhoux (2000) 126.

(133) Cf. also Lys. 744 τέξομαι vs. Thesm. 509 τέξειν.

(134) e.g. vesp. 37; Ran. 122, 340/1; cf. Wackernagel (1926) 122; Schwyzer- Debrunner (1950) 224; MacDowell (1971) 132.

(135) For the perfect (and pluperfect) see the list in Kontos (1898) 287–304, with many examples also from the tragedians, Herodotus, and Thucydides.

(136) See Basset (1979) 132–6, and Duhoux (2000) 161, with statistics: Aristophanes 18 μέλλω + fut. inf. vs. 41 μέλλω + pres. inf. vs. 2 μέλλω + aor. inf. (cf. Aeschylus 6–3-1; Sophocles 11–12-0; Euripides 29–34-13; Herodotus 74–36-2; Thucydides 77–41-5; Isocrates 67–37-0; Demosthenes 106–90-2 or 3).

(137) Ach. 1099, 1101, 1122 (where οἷσε is misleadingly counted as an aorist-stem imperative by Nickau (1993) 163); Ran. 482; cf. Alexis fr. 125; Anaxippus fr. 6; Leumann (1953) 211 n. 2.

(138) See Moorhouse (1946).

(139) e.g. Ach. 166; Nub. 367, 505; Ran. 202; cf. Kühner-Gerth (1898/1904) ii/2. 222–3; Schwyzer-Debrunner (1950) 293; Cooper (1998) i. 653–6.

(140) Cf. Ch. 6.7; Weber (1884) 120–1; Poultney (1963) 369–73; Duhoux (2000) 231–40.

(141) Eq. 669, 893, 1182, 1393; Nub. 539; vesp. 70, 313, 1028; Pax 226; Aυ. 73; Lys.373 (contrast 374); Thesm. 944; Ran. 1419; Eccl. 117.

(142) vesp. no; Aυ. 1524; Ran. 24; cf. Eq. 935; Ran. 766 (but not Eccl. 881, 917, where the optative may be potential; Thesm. 588 depending on 585 ἀναπέμψαι; Pax413 and Aυ. 1338 with attraction of mood).

(143) Bers (1984) 135; cf. Hale (1893) 180–3, 202; Slotty (1915) 139–42; Wackernagel (1926) 236–7: vesp. 472; Aυ. 180?; Lys. 839?; Thesm. 872 (paratragic); Plut. 374?, 438?. In Eq. 1057, Ran. 574, and Eccl. 898 the ἄν of the preceding potential optative may ‘count’.

(144) Eq. 698, aped in Eq. 700; Pax 450; Aυ. 928–30: all can of course be ‘corrected’; cf. Sobolewski (1891) 16–19; Bers (1984) 158–61.

(145) Eq. 805; Lys. 580–1, 1005?; Ran. 1281; Eccl. 629, 688, 752.

(146) Cf. in general Zilliacus (1946); for rhetorical questions Dittmar (1933) 37–41 (more than 80 examples in Aristophanes).

(147) e.g. Nub. 1345, 1397, 1494; Pax 1305; Lys. 315, 839; Ran. 590; Eccl. 514.

(148) Eq. 1161; vesp. 726; Ran. 1401, 1467; Eccl. 132.

(149) often in oracles: Eq. 1039; Aυ. 971, 973, 975; elsewhere (often after a full imperative): Ach. 257; Eq. 1187; Nub. 850, 1080; vesp. 386, 1216; Pax 1153; Lys.536; Thesm. 157; Ran. 133, 169; Eccl. 1107, 1111, 1146?; contrast the ‘optatival’ infinitive (Ch. 2.9) and cf. Dittmar (1933) 92; Bers (1984) 168–82.

(150) e.g. Ach. 204; vesp. 422; Pax 301, 555; Plut. 1196; cf. Coulon (1908) 193–4; Wendel (1929) 25; Svennung (1958) 220–1.

(151) The origin of this aorist imperative may be gerundival: Kretschmer (1919). Hence, it is not used with the imperatival negation μή (Thesm. 870 is a quotation of an irregular use by Sophocles, fr. 493 Radt; cf. Thugenides fr. 4).

(152) See further Nub. 268, 819; vesp. 835; Av. 7–8; Ran. 741; Eccl. 788–90; Plut.593; cf. A. R. Anderson (1914) 74–5; Dittmar (1933) 91–2; Bers (1984) 183–6.

(153) Nub. 1252; Pax 857; Av. 1225 (conjectural); Eccl. 350; cf. Thesm. 34; also Nub. 722 ὀλίγου (sc. δεῖν).

(154) Denniston (1950) pp. lxx-lxxii; for a detailed discussion of various colloquial particles in Aristophanes see López Eire (1996a) 119–33.

(155) Denniston (1950) p. lxxv, 282 n. 1, 568; on τοίνυν Lopez Eire (1996a) 52–5.

(156) Duhoux (1997), correcting e.g. Denniston (1950) pp. lxxii-lxxiii; Schwyzer-Debrunner (1950) 556.

(157) Figures based on Lasso de la vega (1952) 320–5.

(158) But of course they are frequent in gnomic writers (e.g. Hesiod) or tragic cho ruses: Guiraud (1962) 158; Bers (1984) 195–6.

(159) e.g. Eq. 625–82; Nub. 709–15, 1373–6; vesp. 605–12; Pax 605–48; Av.494-503; Lys. 784–92; Thesm. 476–89; Ran. 1192–5; Plut. 653–747; cf. Trenkner (1960) 4–5, Dover (1997) 76, and Ch. 6.8.1 above on paratactic conditional clauses.

(160) Cf. Nub. 409, 624; Av. 536, 674; Lys. 560; Denniston (1950) 308–9; Trenkner (i960), esp. 8, 12–13, and 52–3.

(161) Cf. Kurzová (1968) 64–6; Crespo (1984) 7–10; Delaunois (1988) 143; Cooper (1998) ii. 1040–2.

(162) Fournier (1946) 149; cf. Duhoux (2000) 265.

(163) Monteil (1963) 356, and 399 for the statistics; cf. Fournier (1946) 155.

(164) Monteil (1963) 145, 150, 154, who understands passages like Thesm. 619 or Plut. 941 and 998 as direct questions. In ‘answering questions’ (τί ποιεῖς;—ὅ τι ποιῶ;/τί ποιῶ; ‘What are you doing?—What am I doing? [I am x-ing]’) both interrogative pronouns are used but the direct one less often (Nub. 664; Pax 847; Av.608, 1234; Ran. 1424; Eccl. 761 vs. e.g. Ach. 595; Eq. 128, 1073; Nub. 677, 690, 1495; vesp. 48; Pax 701; Av. 164, 299; Ran. 198).

(165) e.g. Pax 21, 688; Thesm. 801; Plut. 1171.

(166) See in general Sobolewski (1891) 157–79.

(167) Monteil (1963) 135–6: Aeschylus o ὅς : 3 ὅστις; Sophocles 14 : 10; Euripides 13 : 47; Aristophanes 7 : 44; Lysias 17 : 42; Herodotus 41 : 24. Note ὅςτις in a non- defining relative clause at Plut. 13: López Eire (1991) 37.

(168) Statistics in Sobolewski (1891) 177. Relative ἔνθεν and ἔνθα occur only in lyric passages: Av. 748, 1485, 1556; Lys. 625; Thesm. 1046. Note that Aristophanes uses local ᾗ/ὅπῃ) and local ἵνα(Pax 900; Av. 153, 822; Ran. 778), which are both extremely uncommon in Lysias. very rarely (Ach. 364; vesp. 1270), except in dialect passages (e.g. Ach. 730; Lys. 84), ᾗπερ/ᾇ(περ) can replace ὤςπερ. Temporal ὅπως (Nub. 60) and ὅπως for ὡς (Thesm. 1106, paratragic) are isolated in Aristophanes.

(169) See Weber (1884) 127, 132; Weber (1885) 9, 44–5, 55, 74–5; cf. W. W. Goodwin (1897) 398; Monteil (1963) 402; for the inscriptions Meisterhans- Schwyzer (1900) 253–5 and now the index of IG i3; Amiguès (1977) 99 considers only final clauses with subjunctive.

(170) vesp. 162; Av. 1239 (paratragic); Lys. 1093–4; Eccl. 29, 997; Plut. 622. Lysias has only ἵνα μή, not μή alone: Wackernagel (1928) 276.

(171) vesp. 527/8; Pax 310, 432; Lys. 384?, 1094; Thesm. 431, 653; Ran. 1121; Eccl.783, 997; cf. Eccl. 488, 495 with μή, see further Ch. 6.8.1; Weber (1884) 122; W. W. Goodwin (1897) 115–16; Amiguès (1977) 79–82; Duhoux (2000) 457–8.

(172) Ach. 1060; Eq. 81?, 918, 926; Nub. 739; Eccl. 624; W. W. Goodwin (1897) 125–6.

(173) Amiguès (1977) 69; cf. Weber (1884) 124–5; W. W. Goodwin (1897) 94; Kalén (1941)98-137.

(174) Cf. Kurzová (1968) 49; Delaunois (1988) 148–9; Duhoux (2000) 274–5.

(175) Statistics based on the figures in Monteil (1963) 350.

(176) Sobolewski (1891) 13–120.

(177) Monteil (1963) 401; he regards (p. 358) causal ὡς as a recent development originating in colloquial Attic, but this does not explain its use in Aeschylus and Sophocles and its almost complete absence from Lysias.

(178) Cf. Dunbar (1995) 265; statistics after Sobolewski (1891) 120–57, who further discusses the different types of temporal clauses.

(179) López Eire (1986), esp. 249–57, and (1991). López Eire does not always take into account how common a feature is in other 5th-cent. literature.

(180) Cf. Willi (2002b) 112–25; personally I am most familiar with the Fasnacht (= carnival) in Basel (Switzerland) where satirical texts are composed in an extremely conservative (but colloquial) version of the local dialect, which only very few speakers would use with similar ‘purity’ in other contexts (cf. Universitat Basel press release of 19 April 2002, referring to research conducted by a team of students under the direction of Prof. Heinrich Löffler).