ἀμφορέα and other ευ-stem accusatives
(Moer. α 12, Moer. ι 4, Moer. ι 18, Philemo [Laur.] 355)

D. General commentary

Two entries in Moeris' Atticist lexicon (A.1, A.2) and a similar entry in that of Philemon (A.4) discuss the prosody of the accusativeAccusative singular of nouns (and proper names) in -εύς. Another entry from Moeris (A.3) concerns the prosodyProsody of the accusative plural of such nouns, which presents similar problems. These Atticist sources prescribe in both cases the ending with a long alpha (-έᾱ, -έᾱς), remarking that the forms with a short alpha (-έᾰ, -έᾰς) are non-Attic. The same prescription is found in two entries in Photius that Alpers identified as fragments of Orus’ lexicon (B.4, B.5).

While the ultimate origin of nouns in -εύς remains a matter of some controversy (see Perpillou 1973 for a classic treatment and Meissner 2016 for a recent summary), the accusative endings sing. *-ēu̯-m̥, pl. *-ēu̯-n̥s can be safely reconstructed for Common Greek (see Perpillou 1973, 64 and 70–1). In Attic and (less consistently) in Ionic, the inherited endings -ῆα, -ῆας (still preserved in the Homeric language) underwent the process traditionally labelled ‘quantitative metathesis’: the likeliest actual phonetic development was elucidated by Méndez Dosuna (1993) in a seminal article, developing an insight by Schwyzer (1939, 245–6, 575; see also K–B vol. 1, 44; Haug 2002, 107–44; Vessella 2018, 144–6). In the sequences ηᾰ and ηο, the first vowel became non-syllabic and the second simultaneously underwent compensatory lengtheningCompensatory lengthening: thus, the phonetic outcome of quantitative metathesis was ε̯ᾱ (and ε̯ω) with synizesisSynizesis, but a disyllabic scansion arose by secondary ‘distraction’. Indeed, in the oldest metrical texts (Homer, Ionic lyric, Ionic verse inscriptions) the outcomes of quantitative metathesis are predominantly scanned with synizesis, while the frequency of the disyllabic scansion increases with time. A related development, acknowledged by some erudite sources (B.1, B.2, B.7; see also [Arcad.] 283.5–6; Sophronius Grammaticus GG 4,2.390.36–8; Choerob. in Theodos. GG 4,1.238.1–6; Eust. in Il. 1.429.14–7), is the apparent contractionContraction of -έᾱ(ς) to -ᾶ(ς) (and of -έως to -ῶς, -έων to -ῶν in the gen. sing. and plur., respectively) that took place in Attic when the suffix -εύς was preceded by another vowel or diphthong, especially (α)ι but also ο, as in χοᾶ, contracted from χοέα, the acc. sing. of χοεύς ‘pitcher’; the contracted spelling was common in Attic inscriptions until 300 BCE, after which it is found only in the toponym Πειραιᾶ (see Threatte 1996, 248 and 252–5). In light of the current understanding of quantitative metathesis as synizesis plus compensatory lengthening, this apparent contraction may be interpreted as the merger of non-syllabic /e/ with the preceding /i/, i.e. ἁλιᾶ [halijâː] (from earlier ἁλιέα [halie̯âː]) vs. βασιλέα [basile̯âː] (see Vessella 2018, 145 n. 59). In the accusative of eu-stems, most non-Attic dialects, including later Ionic, show the endings -έᾰ, -έᾰς with shortening of /ē/ in hiatus, although, for Ionic at least, an analogicalAnalogy origin (from metathesised -έᾱ(ς) under the influence of the usual athematic endings -ᾰ/-ᾰς) may not be excluded.

All the above-described developments ultimately produced four different endings: (1) -εα with synizesis, (2) contracted -ᾶ, (3) disyllabic -εα with [aː], and (4) disyllabic -εα with [a]. The more common outcomes in classical Attic are (1) or (3) (for the latter, see, e.g., Aesch. Eum. 271). In most dialects other than Attic, (4) is the norm, but this prosody was also occasionally adopted in Attic tragedy (albeit only in lyric sections and only with the noun φονεύςφονεύς ‘murderer’: see C.4 and the evidence collected and discussed by La Roche 1897, 2–4; Arnott 1996, 334). In dialogues, however, tragic writers tended to use the long prosody (see, e.g., C.1, C.2, C.3), and the same is true for Old Comedy (see C.5). Meanwhile, 4th- and 3rd-century comedy attests all four outcomes. Excluding metrically ambiguous cases, synizesis is attested in Antiph. fr. 188.17 (ἁλιέας); contracted -ᾶ is attested in Anaxandr. fr. 33.1, Epin. fr. 2.8, Eub. fr. 80.4, Men. fr. 442 (χοᾶ), Alex. fr. 15.19, Damox. fr. 1.3, Epin. fr. 2.5 (χοᾶς), Men. Epitr. 752 (Πειραιᾶ); [aː] is attested in Timocl. fr. 18.6, 2x in Timocl. fr. 19.3, Men. Epitr. 326, Men. fr. 508.2, Anaxandr. fr. 40.10; [a] is attested in Theophilus (C.6), Euphron (C.7), Antiphanes (C.8), and Philemon (C.9). It has been suggested that in some of these passages, the use of the short prosody might have been intended to characterise non-Attic speech (see F.2).

The many surviving entries concerning the accusatives of eu-stems demonstrate that Atticist lexicographers, especially Moeris, were keen to recommend the long prosody; Philemon here appears to agree with Moeris, as he often does (see Hansen 1998, 40–2; on Philemon’s interest for issues of vowel length see Brown 2008, 224–5). It may seem surprising that the Atticist lexicographers were so concerned for forms such as ἱππέᾰς, which were likely quite marginal in the koine (Vessella 2018, 146): the common endings in Attic and the koine from the end of the 4th century BCE are -ῆς/-εῖς, taken from the nominative plural (see Threatte 1996, 247–8); purists also reacted to this change: see Antiatt. γ 18Antiatt. γ 18 in defence of -έας against -εῖς as the proper acc. pl. ending (see also C.8). The evidence for the short prosody provided by later comedy (see C.6, C.7, C.8, C.9) suggests that Atticists were likely reacting to those who relied on such attestations to defend the prosody with short [a] as good Attic. The correct prosody of the accusative of eu-stems is also discussed in grammatical sources, which are unanimous in considering the ᾱ-forms to be the correct Attic ending (see B.1, B.3, and cf. La Roche 1897, 1).

E. Byzantine and Modern Greek commentary

In the medieval period, the eu-stems were no longer productive and were gradually eliminated from the spoken language, partly as a consequence of the phonetic development /eus/ > /efs/. As in other 3rd-declension nouns, precisely the accusative singular -έα (where the length of α had, of course, become irrelevant after the loss of contrastive vowel quantity) served as a pivot for the analogical transferral of such nouns to a declension in α: a nom. sing. -έας and a gen. sing. -έα were formed, while the acc. sing. itself could take a -ν in analogy to the a-stems, causing a partial merger with inherited forms such as Ἀνδρέας and βορέας (see CGMEMG vol. 2, 401–22). These endings additionally tended to undergo synizesis with stress shift on the second vowel, as demonstrated by spellings such as βασιλεάς or βασιλιάς for nom. sing. βασιλέας ‘king’; however, this is a distinct phenomenon from the synizesis that originally accompanied quantitative metathesis in Ancient Greek, being instead part of the widespread Medieval and Early Modern Greek loss of syllabicity of /e/ and /i/ before another vowel (often concealed by conservative spellings). Nevertheless, forms preserving the classical endings are found throughout the Byzantine period, particularly in loftier words: the nom. sing. βασιλεύς and voc. sing. βασιλεῦ, for instance, often also occur in lower-register texts, ‘probably because [they] sounded more elevated and appropriate for kings and emperors’ (CGMEMG vol. 2, 407; see also Kriaras, LKNE s.v. βασιλεύς). In the accusative plural, the ending -εῖς that had begun to replace -έας already in the 4th century BCE remained ‘by far the most common’ in the Byzantine period. The archaising -έας still appears in mixed- and higher-register texts, while the analogical endings -ιάδες/-εάδες (15th century) and -έες (only Irodis 5.179, 17th century?) are attested late.

In Modern Greek, the remnants of old eu-stems are inflected according to the innovative declension in -α, but the conservative forms may survive in archaising language and in set phrases. Thus, for instance, the usual word for ‘king’ is βασιλιάς /vasiˈʎas/ (besides the more conservative variant βασιλεάς /vasiˈleas/ for non-metaphorical meanings: see LKN s.v.), while the nom. sing. βασιλεύς /vasiˈlefs/ and gen. sing. βασιλέως, pl. -έων may be found in idioms such as ο βασιλεύς των βασιλέων ‘the king of kings’ (i.e., Christ), dating back to Hellenistic Greek, or βασιλικότερος του βασιλέως ‘more royal than the king’ (a modern creation with archaising morphology, calqued on French plus royaliste que le roi).

F. Commentary on individual texts and occurrences

(1)    Phot. α 3111 (B.7)

This lemma, transmitted by cod. z of Photius’ lexicon and in the Suda, concerns the accusative singular ending of masculine names in -ευς. It observes that the ending -έα is only found with a preceding consonant, since when preceded by a vowel, it is contracted to -ᾶ (cf. the Herodianic doctrine in B.1, B.2, and other anonymous treatises Περὶ διχρόνων that are more or less dependent on Herodian: cf. also Anon. Περὶ διχρόνων 357.23–5 Egenolff; Anon. Περὶ διχρόνων 8.16 Sandri) and that the α in -έα is long in Attic but short in Ionic (and sometimes in Attic tragedy). A comparison is then drawn with names in -ης (i.e., masculine s-stems), which similarly have an accusative singular with short α in Ionic. To exemplify this point, a quotation from Aristophanes’ lost play Farmers is introduced. Unfortunately, the text of the fragment is seriously corrupt, although most editors, following Gaisford’s emendation, read the name Μεγακλέας (plur.) or Μεγακλέα (sing.) ‘Megacles/-eses’, while †καχέτας† may conceal another proper name (see Bagordo 2022, 22–3 on the textual difficulties and the various emendations proposed, observing that in the context of this lemma, a singular would be expected, although it is true that the gloss is concerned with the eu-stem accusatives in general). The accusative singular of s-stem compound names in -κλῆς (< -κλέης), indeed, was the object of Atticist prescriptions: cf. Phryn. Ecl. 127Phryn. Ecl. 127, recommending the ending -έᾱ (to which Thom.Mag. 171.14Thom.Mag. 171.14 adds the contracted -ῆ as an admissible form) against the analogical -ῆν. While the distribution ‘Attic -έᾱ : Ionic -έᾰ’ in such names is superficially the same as in eu-stems, its origins differ somewhat: the Attic ending -κλέᾱ is contracted from -κλέεα (< *-κλέϝεα < *-kleu̯-es-m̥), while Ionic has -κλέεα > -κλέᾰ by hyphaeresisHyphaeresis (on which, see now in detail Nussbaum 2017).

(2)    Theophil. fr. 1 (C.6)

This fragment from the Ἀπόδημοι (‘Emigrants’, a form contrasted to Attic ἔκδημος by Moer. ε 17Moer. ε 17) of the Middle Comedy poet Theophilus (4th century BCE), quoted in a scholium to Dionysius Thrax (GG 1,3.159.23–6), contains the monologue of a slave of apparently non-Greek origin who appears reluctant to leave his master and benefactor behind, perhaps after having been freed by him. The paratragicParody tone of the speech is also apparent from the choice of the word τροφέατροφεύς ‘foster-father’, which does not occur elsewhere in comedy but has some occurrences in tragedy (e.g., Aesch. Ag. 729, Soph. Ph. 344). Here the word must be scanned with a short final α: Papachrysostomou (2008, 251) thinks that the use of ‘the non-Attic form τροφέα […] in the mouth of a non-Attic speaker within a play that has for a title a not particularly Attic word […] is all too much to be a coincidence’, and compares the use of the allegedly non-Attic ψυγέᾰψυγεύς ‘cooler’ in Euphro fr. 3 (C.7), in which the speaker contrasts Attic and non-Attic words. Thus, Theophilus would have combined a lexical form that was untypical for Attic with a non-Attic ending to characterise the speech of a foreigner who, according to his own testimony, has learnt Greek as a second language. While this is an intriguing suggestion, it remains speculative: note the slave’s speech includes no other traits that are typical of non-native Greek (such as that of the Scythian archer in Aristophanes’ Thesmophoriazusae, on which, see Willi 2003, 198–225) nor any other obviously non-Attic traits. It is possible that τροφέα, like, e.g., δρᾶν at v. 1, was simply chosen to impart a tragic flavour. In Euphron’s passage (which does make a metalinguistic point), the term ψυγεύς is contrasted with ψυκτηρίας, a different formation from the same root: the focus is on non-Attic word-formation (see Arnott 1996, 193 on ψυγέα in Alex. fr. 65), rather than non-Attic prosody, although the other word pairs contrasted (σεῦτλον vs. τευτλίον, φακῆ vs. φακέα) indeed show different phonological treatments. Such comparisons between Attic and non-Attic vocabulary were topical in later comedy: for σεῦτλον vs. τεῦτλον/τευτλίον cf. Alex. fr. 146.5–7, Diph. fr. 46, Luc. Iud.Voc. 10, and see Arnott (1996, 433–4).