ἰχθῦς, δρῦς
(Phryn. PS 77.14–5, Philemo [Laur.] 359)

D. General commentary

These entries in Phrynichus’ Praeparatio sophistica (A.1) and in the lexicon of Philemon (A.2) discuss the accusative and nominative plural of the nouns ἰχθύς (~ ἰχθῦς: on the accent of the nom. sing. see Dieu 2022, 343) ‘fish’ and δρῦς ‘tree, esp. oak’, respectively. Both nouns belong to the non-ablauting u-stems, a category that resulted from the merger of the inherited stems in *-ŭ- and in *-ū- < *-uH- and that also attracted several nouns that originally belonging to other stem-classes, such as μῦς ‘mouse’, originally an s-stem (cf. Latin mus muris). The earliest and best attested accusative plural of this nominal class is -ῦς (< *-u-ns), which originally belonged to the ŭ-stems, while the -uH-stem ending must have been *-u(H)n̥s > -υας, cf. Vedic -uvas (see Chantraine 1961, 93–4; Beekes 1973, 229–41; Sihler 1995, 327–8; Martínez García 1996; on the Indo-European prehistory of u-stem nouns see in detail Neri 2003, 45–114). An ending -υας – whether a survival of the inherited -uH-stem ending, or an independent analogical innovation on the model of consonant stems – is indeed attested from the Homeric poems before becoming more common in late Attic and in the koine. In Homer, the acc. pl. typically has the ending -ῡς, though -υας also sometimes occurs and is metrically guaranteed in several cases (e.g. C.1), although in two instances (Od. 9.389 and 22.384 = C.2), -ῡς may be substituted (Schwyzer 1939, 571; Chantraine 1958, 222). The textual tradition of Herodotus has both endings, with -ῡς in the majority; Rosén (1962, 78) proposed that the distribution reflected a genuine sandhi phenomenon in Herodotean Ionic, with -ῡς being the prevocalic and prepausal allomorph and -υας the preconsonantal one. However, this theory is highly tenuous: the mixture of forms might also reflect, for instance, the influence of the Homeric and poetic traditions. In older Attic texts, the ending -ῡς predominates (see e.g. C.4, C.5 for Aristophanes), but -υας, which enjoyed the analogical support of consonant stems, gained traction in Post-classical Greek. An early instance of the innovative ending occurs already in a Sophoclean fragment (C.3) preserved by Hesychius (B.3) and then in Posidippus (first half of the 3rd century BCE), a poet of New Comedy (C.9). Phrynichus (A.1) duly indicates the accusative plural ἰχθῦς as ‘more Attic’ than ἰχθύας.

In parallel to the remodelling of the accusative plural, in post-classical times u-stem nouns also underwent remodelling of the nominative plural ending. The older ending -υες, already attested in Mycenaean (ta-ra-nu-we /tʰrā́nuwes/ ‘(foot)stools’), which is the only form in Homer and still the norm in 4th-century literary Attic, is occasionally replaced by -ῦς, thought to be modelled on the acc. pl. itself (K–B vol. 1, 439; Schwyzer 1939, 564; Martínez García 1996, 259–63). This innovative Attic ending first appears in Middle and New ComedyComedy (see Gomme, Sandbach 1973, 555; Sommerstein 2013, 134; Cartlidge 2014, 209): for instance, nom. pl. μῦς ‘mice’ appears in Antiph. fr. 191.1 (C.6); nom. pl. ἰχθῦς ‘fish’ in Eub. fr. 108.3 (C.7), Alex. fr. 47.2, fr. 263.9, Antiph. fr. 233.3, Men. Sam. 98 (C.8). Later comedy thus aligns with a tendency that is already discernible in Attic documentary inscriptions at the end of the 5th century: see Threatte (1996, 219–20), who discusses the occurrence of the nominative plural στάχῡςστάχυς ‘ear of corn’ in place of στάχυες. In addition to comedy, the innovative ending is also used in Attic prose by Xenophon for a technical term (ἄρκυς ‘hunter’s nets’) in a technical work (Cyn. 2.4.3, 2.9.2, 6.2.5, 10.2.1, 10.19.1; see K–B vol. 1, 439 and Schwyzer 1939, 564). This innovation did not escape the attention of Atticist lexicography either, as suggested by the entry in cod. L of Philemon’s lexicon (A.2) dealing with the forms δρῦς and δρύες, although according to Osann’s text δρύας, the entry would be in contrast with the accusative forms (see the discussion in F.1).

Atticising writers generally follow the Attic norm while also admitting the innovative accusative forms, with some individual differences: Aelius Aristides is the most rigorous in employing the acc. pl. in -ῦς (see Schmid, Atticismus vol. 4, 581); Lucian typically uses accusatives in -ῦς but uses -ύας (ἰχθύας, σύας, ὀφρύας) in works imitating Herodotean prose (e.g., The Syrian Goddess, Astrology), suggesting that this ending may have been perceived as typical of literary IonicIonic in contrast to the ‘Attic’ -ῦς (in fact, the text of Herodotus shows a mix of the two endings: see above); Aelian has acc. pl. ἰχθύας, μύας but also ἰχθῦς, μῦς as well as βότρυς, σῦς, ὗς, στάχυς (Schmid, Atticismus vol. 3, 22, who also reports the coexistence of μύας and μῦς in Babrius); in Philostratus, accusatives in -ῦς are considerably more frequent than those in -ύας (Schmid, Atticismus vol. 4, 19). The situation is similar with the nominative plural: Lucian, for instance, only uses ἰχθύες (5x), στάχυες (2x), σύες (3x) and never the corresponding -ῡς-forms; Aelian, meanwhile, uses the nom. pl. ἰχθῦς and σῦς/ὗς side by side with ἰχθύες and σύες/ὕες.

The uncertainty of 2nd-century-CE speakers regarding the correct declension of ἰχθύς may be partly explained by the fact that use of the word itself in popular speech was declining, to be replaced by ὀψάριονὀψάριον, the ancestor of Modern Greek ψάρι, as attested by Phryn. PS 81.13Phryn. PS 81.13: ὀψάριον τὸ ὄψον, οὐχὶ τοὺς ἰχθῦς. οἱ δὲ νῦν τοὺς ἰχθῦς <οὕτω> λέγουσιν ‘ὀψάριον: [It means] ‘relish’, not fish. But contemporary speakers call fish ’.

The situation outlined above is reflective of a wider trend in the historical development of Greek – namely, the gradual elimination of masculine and feminine nouns of the 3rd declension that largely merged with the 1st declension, partly due to the confusion between nominative and accusative plural endings (Horrocks 2010, 117). As a marginal category with few members and already unproductive since antiquity, the u-stem nouns were an early target for analogical remodelling (or, ultimately, lexical substitution), and they did not survive as a class into Modern Greek, with the exception of some learned forms (see E.).

Ancient grammarians offered different explanations for the relationship between the various forms. According to the testimony of the Iliad scholia (B.1), John Philoponus (B.4) and Choeroboscus (B.6), Herodian claimed that the acc.pl. -ῦς had the same accentuationAccent as the nom. pl. -ῦς and explained both as the outcome of contractionContraction, the former from -ύας and the latter from -ύες. Theodosius (B.2) presented -υας as the regular form, according to the rule that athematic nouns with a nom. pl. -ες have a corresponding acc. pl. in -ας. Later grammarians, such as Sophronius (B.7) and Choeroboscus himself (B.5), objected to the idea that the acc. pl. -ῦς was the result of contraction and instead explained it as simply homophonous with the nom. pl. -ῦς, according to Theodosius’ rule. The correspondence between ἰχθῦς and ἰχθύας on the one hand and that between ἰχθύες and ἰχθῦς on the other may have given speakers the sense of a contraction in the -ῦς-endings: in turn, this fact – together with the influence of nom./acc. pl. -εῖς – may account for the circumflex accentuation of these endings, which may be an Atticism in the tradition of Homer’s text (see Dieu 2022, 497, pointing to scholiastic evidence for an old accentuation -ῡ́ς in the accusative).

E. Byzantine and Modern Greek commentary

Both nouns treated in these entries share a similar fate in Medieval Greek. Both survive exclusively in the formal register and are relatively uncommon; moreover, both undergo various attempts at morphological regularisation. In the case of ἰχθύς, ‘the ancient variant ἰχθῦς for acc. pl. is much rarer than ἰχθύας’ (CGMEMG vol. 2, 453); in the case of δρῦς, among various analogical innovations in other case-forms, the classical nom. pl. δρύες is found in the function of accusative, and the noun may even be treated as undeclinable (CGMEMG vol. 2, 453 and 589). In Modern Greek, the learned term δρύς survives in the meaning ‘oak (wood)’ in addition to the popular term βελανιδιά (from βαλανίδι ‘acorn’, a diminutive of ancient βάλανον); similarly, ιχθύς is the katharevousa term for ‘fish’ (supplanted by ψάρι in demotic Greek) and, in the plural οι Ιχθύες, the name of the constellation of Pisces (see LKN s.vv.). The innovative term was originally a diminutive of ὄψον ‘(side) dish, relish, delicacy’, which progressively restricted its meaning to fish, the delicacy par excellence in ancient Greece (on the semantic evolution of ὄψον, see Kalitsounakis 1926; Shipp 1979, 427‒9). ὀψάριον followed the development of the base word, being attested in Attic comedy still with the meaning ‘side dish’ in Aristophanes (fr. 45) and Alexis (fr. 177) but otherwise mostly referring to fish already (cf. Ar. fr. 45, Pl.Com. 95, Pherecr. fr. 27, Philem. fr. 32, etc.): these attestations were collected and discussed by Ath. 9.385d–fAth. 9.385d–f, with the aim of identifying classical antecedents for contemporary usage. Interestingly, in the New TestamentNew Testament, ἰχθῦς is used to refer to living fish, while ὀψάριον is used to denote fish as food; Shipp (1979, 427) remarks that this usage is already evident in Menander (Car. 1.2, Eph. 151.2), where ὀψάριον is the object of forms of λαμβάνω ‘catch’.

F. Commentary on individual texts and occurrences

(1)    Philemo (Laur.) 359 (= A.2)

The actual prescription contained in this (highly abridged) lemma from Philemon’s lexicon is not immediately apparent. The form δρῦς (which may be either the classical accusative plural or the innovative nominative plural) is contrasted with a form that is transmitted in cod. L as δρύες: Osann (1821, 300) printed it as δρύας, possibly simply resulting from a misreading of L, while Cohn (1898, 359) tacitly restored the manuscript reading. According to the transmitted text, this is a discussion of the nominative forms, contrasting the innovative δρῦς with the conservative δρύες; if δρύας is read, it would become a discussion of the accusative forms, contrasting conservative δρῦς with innovative δρύας, much as Phrynichus chose ἰχθῦς/ἰχθύας to comment on the same phenomenon. Given that both the conservative and the innovative forms of the nominative and the accusative plural are attested for this word in literary texts, both interpretations are defensible. In a clear majority of cases (53 vs. 11), Philemon gives the correct form first (see Brown 2008, 211–5): as such, although it cannot be ruled out that this entry lemmatises an innovative nom. pl., it is more likely that it lemmatises the correct Attic acc. pl. A third possibility is that it admits both forms as correct: indeed, Philemon occasionally gives two alternative forms as acceptable (see Brown 2008, 223), although this double possibility is typically indicated with the use of ἀμφότερα/-ως or ἑκάτερα/-ως. The verbal form φαμέν ‘we say’ is not found anywhere else in the transmitted versions of Philemon’s lexicon but is sometimes used by Phrynichus (5x) and Pollux (2x), in which it does not appear to have a precise technical meaning: it may introduce a prescription or a description of correct usage contrasted with an incorrect one (Phryn. Ecl. 253Phryn. Ecl. 253, 285Phryn. Ecl. 285, 403Phryn. Ecl. 403, 419Phryn. Ecl. 419) as well as a reference to common usage (Phryn. PS fr. *245Phryn. PS fr. *245 [= Phot. α 2252] ὡς ἐν τῇ συνηθείᾳ φαμέν ‘as we say in the common language’). In Philemon’s entry, the presence of the conjunction καί ‘also’ suggests that both forms discussed (whichever they actually were) were at least deemed acceptable: this accords with the mixed distribution of conservative and innovative nom. pl. and acc. pl. forms in Atticising writers, as outlined above.