γυναί
(Antiatt. γ 1)

D. General commentary

This lemma in the Antiatticist (A.1) concerns several analogical forms of the noun γυνή (‘woman, wife’): nom. pl. γυναί and acc. sing. γυνήν in place of γυναῖκες and γυναῖκα, respectively. As is well known, this irregular third-declension noun is inflected as a velar stem γυναικ- outside the nom. sing., but analogical forms of the other cases following the regular ā-stem declension are occasionally found in Attic comedy and attracted the interest of grammarians and lexicographers.

From a historical-linguistic perspective, such forms are innovations, as a look at the history of γυνή will show. This noun goes back to the consonant stem *gʷén-h₂-s, *gʷn-éh₂-s ‘woman’, which underwent various morphological remodelings across the Indo-European languages (NIL 177–85). Greek agrees with two closely related languages – Phrygian and Armenian – in showing an extension -ai(k)-: PhrygianPhrygian is closer to Greek, with a stem knai(k)- throughout the paradigm (see Obrador-Cursach 2020, 273–4), while ArmenianArmenian has a stem kin-/kan- in the singular but kanya- in the nominative and accusative plural (see Clackson 1994, 136–7; Szemerényi 1960, 13–23 doubted that the stem extensions in Greek and Armenian were related). These dialectal attestations taken together with the Phrygian and Armenian evidence indicate that all Greek dialects inherited the extended stem: BoeotianBoeotian, for instance, has the nom. sing. βανά (Corinn. fr. 664a, Hsch. β 184) and acc. pl. βανῆκας (Hsch. β 187), neatly corresponding to the Attic forms except, possibly, for the root ablaut (while βανά reflects a zero grade, γυνή may reflect either a full or a zero grade: see Vine 1999, 560–2). The different initial consonants in Bοeotian and Attic strongly indicate that the extension -aik- is older than the dialectal developments of the labiovelars; in principle, however, the Boeotian suffix could have been modified later under Attic influence. The antiquity of the velar stem is further confirmed by the Mycenaean dat. pl. ku-na-ki-si (TH Av 100.2, Fq 200.2) if it denotes /gunaiksi/ = γυναιξί ‘women’, as is more likely (Bernabé 2008, 19–20), and not /kunāgisi/ = κυνᾱγίσι ‘huntresses’.

The most widespread view (see, e.g., Schwyzer 1939, 582–3) is that the nom. sing. Attic-Ionic γυνή – Boeotian βανά – reflects a prehistoric form *gʷ(o)nā́ that, unlike the other instances, did not receive the extension -aik-. Opfermann (2017, 59–61) recently expressed a dissenting view, suggesting that γυνή is itself an innovation (possibly based on the voc. sing. γύναι) that replaced the expected nom. sing. *γυναῖξ but could not oust the stem γυναικ- from the rest of the paradigm. Be that as it may, the pattern of attestation of the ā-stem forms strongly suggests that they were analogical creations (Cartlidge 2016, 19 compares learner errors in English like *childs for children), typical of informal, low-registerRegister speech. In particular, erudite sources attribute such forms to two poets of Old Comedy, Pherecrates (C.1) and Alcaeus (C.2, C.3), and to two poets of New Comedy, Menander (C.4) and Philippides (C.5). The complete absence of ā-stem forms of γυνή from ‘serious’ literary genres and from Attic inscriptions (Threatte 1996, 274) may constitute further proof that they were regarded as substandard. However, their colloquial character made them apt for occasional use in comedy (Urios-Aparisi 1992, 545; Franchini 2020, 51), perhaps also metri causa (Hartwig 2022, 88). It seems more problematic to assume, as did Krumbacher (1885, 533), that they were purely artificial forms coined for comic use.

Given the Antiatticist’s penchant for defending post-classical forms by pointing out their attestations in acceptably ‘Attic’ authors, his mention of two Attic playwrights of Old and New Comedy who employed ā-stem forms of γυνή suggests that such forms were in use during his lifetime, not only in the spoken language but also on the part of some writers. It may be that other contemporary lexicographers had criticised this use and that the Antiatticist was responding to such criticism, even though surviving Atticist lexicography preserves no discussion of the analogical forms of γυνή. Only Sophronius’ commentary on Theodosius’ Canones (B.6) preserves the explicit prescription that the analogical forms, while marginally attested, should not be used.

The Antiatticist entry should be considered against the background of the grammatical debate on heteroclisis: γυνή was among the nominal forms that attracted scholarly attention because some cases of their paradigm seemed to derive from a different stem from the nominative singular. Apollonius Dyscolus (B.1, B.2) theorised that heteroclitic nouns possessed two nominative forms, only one of which was actually attested, while the other served as the derivational basis for those oblique cases that could not be regularly derived from the attested nominative (see F.1). Accordingly, he postulated a nom. sing. γύναιξ as the basis for the forms in γυναικ-. Later authorities accepted and elaborated this explanation, possibly mediated by Herodian (whose doctrine is reconstructed by Lentz in GG 3,1.46.13–23 based on Choeroboscus [B.3], Theognost. Can. 132.30 and Eust. in Od. 1.340.22, 2.275.45) and Theodosius (GG 4,1.14–16, commented in B.4 and B.6). It should be noted that -γύναιξ is in fact marginally attested in compoundingCompounds: Choeroboscus (B.3) mentions ἀγύναιξ (Soph. fr. 4), while Sophronius (B.6) mentions the Homeric and poetic compound καλλιγύναιξ, which, however, is never found in the nominative. Several further compounds with γυνή as the second member follow the athematic declension (as opposed to the more frequent compounds in -γύνης, -γυνος, -γύναιος), but the only attestation of a nominative singular is Sophocles’ ἀγύναιξ: the other instances are the acc. sing. ἡμιγύναικα (AP 6.217.9), ὀρσιγύναικα (fr. adesp. 85.1 PMG) and the nom. pl. πρωτογυναῖκες (Hsch. π 4164), φιλογύναικες (Pl. Smp. 191d7). It is possible that the nom. sing. was avoided due to its aberrant shape with respect to the simplex (see Sommer 1948, 62–3). At any rate, γύναιξ may easily be recovered from the oblique cases on the model of ‘regular’ velar stems, such as θώραξ, θώρακος. The fact that γύναιξ remained unattested may be attributed to its uniqueness among nouns in -αιξ as feminine (B.3) and disyllabic (B.5). Inversely, grammarians sought to explain why γυνή did not produce a complete paradigm like other feminine nouns in -ή: Choeroboscus (B.4) adduces its uniqueness among disyllables in CυCη for having a short υ, although elsewhere (GG 4,2.18.35–19.3) he simply admits that γυνή and ὕδωρ lack a genitive ‘by chance’ (κατὰ τύχην). For this theory, analogical forms like γυναί, γυνῆς, etc., were something of an embarrassment, verifying that γυνή could, in fact, build a complete 1st-declension paradigm.

E. Byzantine and Modern Greek commentary

The difficulties inherent in the irregular inflection of γυνή were solved in various ways in Byzantine Greek. The first and most radical solution was to treat the word as indeclinable. Already in a Ptolemaic papyrusPapyri (P.Grenf. 2.26.14 [= TM 70]; Pathyris, 103 BCE), γυνή occurs where a dative is syntactically expected; however, even if the scribe did not intend to write an ā-stem dative γυνῇ, this may be explained as a parenthetical nominative (Vierros 2012, 166). Another Ptolemaic papyrus (SB 5627.13 = SB 6319.col. ii.45 [= TM 7209]; Magdola, 1st c. BCE) attests the syntagm ἀλλοτρίαν γυνή, with the noun apparently employed as accusative; however, more compelling evidence for its indeclinability is found in its prepositional constructions, such as διὰ γυνή or ὑπὲρ γυνή, attested in papyri from the Roman and Byzantine periods (Mayser, Gramm. vol. 1,2, 30–1; Gignac 1977, 52). The first certain attestation of an analogical ā-stem form in the papyri is the gen. plur. γυ̣ν̣ῶν̣ – interestingly, a form absent from extant literary and erudite sources – in P.Adl. 9.6 (= TM 9) [Pathyris, 104 BCE] (see entry μισογύνης, μισογύναιος).

The Byzantine period simultaneously yields several attestations of inflected ā-stem forms of γυνή (e.g. Digenis Akritis cod. G 6.831: ταύτην οἰκτείρας ὡς γυνήν ‘pitying her as a woman’, Neophytus Inclusus Liber quinquaginta capitulorum 22.1.2: τῆς γυνῆς ἡ ἀπάτη ‘the woman’s deceit’, etc.). It is difficult to determine whether the ā-stem forms found in later periods are survivals of those attested in Attic comedy or independent creations, since the irregular declension of γυνή made it a potential target for analogical renewal throughout the history of the Greek language, and the ending -ή of the nom. sing. could always provide a starting point for the creation of ‘regular’ ā-stem forms, as 1st-declension feminine nouns in -ή of the type φυγή, φυγῆς remained a productive category up to the present day. Krumbacher (1885, 529–35), now followed by CGMEMG (vol. 2, 533, n. 107), claimed that the ā-stem forms found in Attic comedy were artificial creations, while those in Byzantine writers were literary hyper-corrections reacting against the demotic form ἡ γυναῖκα, τῆς γυναίκας with a generalised velar stem. This view may find support in the fact that Medieval ā-stem forms occur only in the singular: the plural only had the stem γυναικ- already in Ancient Greek, and so it could not be perceived as a ‘vulgar’ innovation.

The Modern Greek form is γυναίκα (ILNE 273–4), attested already in Byzantine sources (LBG vol. 1, 332) and built on the accusative singular of the classical paradigm, like many other feminine 3rd-declension nouns that shifted to the first declension thanks to the phonetic confusion between the accusative endings -αν and -α (CGMEMG, vol. 2, 458–9); Gignac (1977, 45) collected numerous examples of acc. sing. γυναῖκαν in Roman papyri. Learned forms, such as γυνή or γυναιξί, survive marginally in set phrases (see LKN s.v. γυνή).

F. Commentary on individual texts and occurrences

(1)    Antiatt. γ 1 (A.1)

The extremely epitomised text of the Antiatticist entry leaves some doubt as to the attribution of the two forms cited and the noun’s grammatical case as it appeared in the texts. Hartwig (2022, 87–8) argues that the lemma’s nominative γυναί likely reflects Philippides’ text, against Bergk who suspected that the second quotation in Choeroboscus (B.4), with no explicit indication of the author and containing the accusative γυνάς, actually came from Philippides. The Antiatticist’s following reference to the use of acc. sing. γυνήν in the Krapataloi of Pherecrates has been identified with the first passage quoted by Choeroboscus, where precisely that form appears. While, as Hartwig argues, there is no reason to doubt that ‘Choeroboscus has omitted Pherecrates’ name when quoting the second passage for the sake of economy’, neither is there any reason to assume that both Pherecrates fragments came from the Krapataloi (Franchini 2020, 49).

(2)    Apoll.Dysc. Synt. 2.19 (= GG 2,2.140.4–9) (B.1)

Following the analogical principle, Apollonius Dyscolus postulates that for heteroclitic nouns and adjectives whose genitive singular is not regularly formed on the nominative singular but for which other oblique cases regularly follow the genitive, there is in fact another nominative whence the genitive singular and the other oblique cases regularly derive. In the case at hand, given that nom. pl. γυναῖκες is regular with respect to gen. sing. γυναικός, there must exist a nominative singular γύναιξ whence the attested genitive is regularly derived. At 2.251 (B.2), on the other hand, Apollonius observes that the theoretically ‘regular’ forms are not in fact attested, while equivalent forms are used in their place. As Lallot (1997 vol. 2, 150 and 89–90) observes, these two positions do not necessarily contradict one another but may be seen as complementary, especially if we assume that the ‘regular’ nominatives at 2.19 are postulated at a somehow more abstract level, regardless of the fact that some may actually have existed (πολλός in Homer) or were created later in the history of the language (Modern Greek γυναίκα, μεγάλος) while others did not (*ὕδας).

(3)    Epim.Hom. γ 25 (B.5)

This entry in the Epimerismi Homerici discusses the different origins of voc. sing. γύναι (from nom. sing. γύναιξ by loss of final ξ), nom. pl. γυναί (not apocopated from γυναῖκες, but from the same ā-stem sporadically found also in other case-forms) and γυναι-, found as first member in some compounds (for which two possibilities are given). The entry quotes three literary attestations of ā-stem forms: one, from Alcaeus Comicus (C.1), is only transmitted here; one, from Pherecrates fr. 96 (C.2), is also transmitted by Choeroboscus (who himself may have been the author of the Epimerismi Homerici: cf. Dyck 1986, 5–7; Dyck 1995, 23–4); the attribution of the other (C.3) warrants a brief discussion. The corrupted manuscript reading μιμηέρμνω appears to suggest the proper noun MimnermusMimnermus, which was restored by Cramer, but the verse – a comic trimeter – and the fact that all other attestations of the analogical forms of γυνή are found in comedy, make this attribution improbable: hence Meineke’s conjecture Μενάνδρῳ and Bergk’s Ἑρμίππῳ, while Nauck thought of the tragedian Mimnermus. The exclamation ὦ Ζεῦ πολυτίμητ’ is typical of comedy, being equally well-attested in both Aristophanes (Eq. 1390, Av. 667, fr. 336.1) and Menander (Peric. 313, Mis. 685, fr. 249). The same line is quoted without the author’s name at Et.Gen. s.v. Συρακοῦς p. 275 Miller. See further the critical apparatus of West (who accepted the fragment among Mimnermus’ dubia) and Kassel and Austin ad loc.