ἥρως, ἥρωες
(Phryn. Ecl. 129)

D. General commentary

An entry in Phrynichus’ Eclogue (A.1) discusses the nominative and accusative plural of ἥρως ‘hero’, prescribing ἥρωες (against ἥρως) for the former and ἥρως for the latter. Interestingly, Phrynichus admits that Aristophanes used the nom. plur. ἥρως on one occasion but cautions that this example should not be followed on the grounds that it is metrically conditioned. The same noun is also treated in Philemon’s lexicon (B.1), which prescribes ἥρω (against ἥρωα) in the acc. sing. but admits both ἥρωες and ἥρως in the nom. plur. (for Philemon’s use of ἑκατέρωςἑκατέρως or ἑκάτερα to label two forms as equally acceptable, see Brown 2008, 216‒7).

The noun ἥρως belonged to a very small inflectional class in Ancient Greek and, moreover, exhibited irregular behaviour from the start. While it is traditionally classed as a stem in *-ōu̯-, like πάτρως ‘paternal uncle’ (cf. Latin patruus) or μήτρως ‘maternal uncle’, not only is -ϝ- practically unattested in alphabetical Greek but the Mycenaean theonym ti-ri-se-ro-e /trishērō(h)ei/ ‘Thrice-Hero’ (dat. sing.) fails to show the expected -u̯-. Nevertheless, it is likely that ἥρως was indeed a diphthongal stemDiphthongal stems    in origin, albeit remodeled as an -ō(s)- stem already in Proto-Greek: see García Ramón (2016, 53‒6) for further details on the derivational history of this word, which in his opinion generalised the vocalism arising in the prehistorical acc. sing. *-ou̯m > *-ōm in accordance with Stang’s Law (see e.g. Weiss 2020, 124). While ἥρως is inflected athematically and imparisyllabically (with acc. sing. in -ωα or elided -ω’, nom. plur. in -ωες, and acc. plur. in -ωας: see LfgrE vol. 2, 938‒41) in Homeric dialect, in classical Attic, it developed a mostly parisyllabic inflection, following the so-called ‘Attic second declension’ of the type λεώς ‘people’ or νεώς ‘temple’, with which it shared a nom sing. in -ως (see Allen 1871, 30; K‒B vol. 1, 455‒6). Threatte (1996, 268‒71) demonstrated that parisyllabic forms, such as gen. sing. ἥρω, are prevalent in Attic inscriptions, although not all cases are equally well-attested (the acc. sing. occurs only once, as -ω, while the acc. plur. is unattested), and pointed out that athematic forms begin to appear after 350 BCE. This trend intensified in later Greek; in Roman and Byzantine papyri, ἥρως ‘is inflected throughout according to the diphthongal stem declension. Attic second declension forms do not occur’ (Gignac 1981, 82).

The peculiar inflection of ἥρως in Attic attracted the interest of grammarians, who were perhaps concerned with the homonymiesHomonymy (ἥρω acc. or gen. sing.; ἥρως acc. or nom. plur.), and of Atticist lexicographers, who were struck by the irregularities vis-à-vis the ‘regular’ Attic inflection, and wished to restore the parisyllabic declension of classical Attic. Moeris, for instance, prescribed ἥρῳ against ἥρωϊ as the dative singular (see entry ἥρῳ). According to Choeroboscus (B.3), the use of nom. plur. ἥρως by Aristophanes was also noticed by Herodian: the verse that Herodian quotes to exemplify this usage, from the lost play Heroes (C.4), may be the locus classicus behind Phrynichus’ lemma. Aristophanes is also quoted in a scholium to the Iliad (B.2) in relation to the use of acc. sing. ἥρων, which is labelled as ‘Attic’: in fact, this is the only attestation of ἥρων in the entire corpus of Attic literature (for a possible poetic attestation, see IG 12,3.1349/863.2 [θε]ιὸν ἥρων [Ἀρτ]εμίδω[ρον] [Thera, ca. 250 BCE], unless elided ἥρων(α) should be read). From an etymological perspective, ἥρων is the regular outcome of the inherited acc. sing. *hērōm < *hērou̯m (just as acc. plur. ἥρως is likely also a regular archaism from *hērōs < *hērou̯ms, cf. Doric acc. plur. βῶς ‘cows’ < *gʷou̯ms); however, it may also be an innovation based on forms such as νεών ‘temple (acc. sing.)’, and ἥρω, the regular outcome from (innovative) ἥρωα. Interestingly, ἥρων was in turn recharacterised as ἥρωνα (first attested in IG 12,4.72.23 [Cos, 240 BCE]; cf. Hom. Ζῆν-α for Ζῆν < *Di̯ēm < *Di̯eu̯m), on the basis of which yet another ‘regularised’ inflection as a nasal stem was created (cf. dat. plur. ἡρώνεσσι in Sophr. fr. 151; see Alonso Déniz 2022, 159, who argues that the coexistence of *hēr-ōn- and *hēr-ō(h)- must have been of sufficient antiquity to serve as an analogical model for the creation of the accusatives *Apoll-ō(h)-a and *Poteidā(h)-ō(h)-a > Ἀπόλλω, Ποσειδῶ next to *Apoll-ōn-a and *Poteidā(h)-ōn-a > Ἀπόλλωνα, Ποσειδῶνα). As for ἥρων, this form is found once in Herodotus (C.3), although elsewhere in his text, the form ἥρωα is transmitted (always before α: 2.143.17, 6.69.15). Therefore, Rutherford (1881, 248) argued that the scholium (B.2) was mistaken in attributing ἥρων to Aristophanes and that the ‘correct’ Attic form was ἥρω (which is, indeed, the form prescribed by Philemon, B.1). However, as Willi remarks (2003, 242 n. 54; see also Willi 2010, 498; Bagordo 2017, 105‒6), these are ‘[i]rregular declension forms’ that ‘belong to marginal and (synchronically) irregular words where people may have been uncertain about the ‘correct’ forms’, as opposed to those non-standard forms that (being marked) ‘have a distinct stylistic function’, such as parody or imitation of non-Attic dialects. Elsewhere, indeed, Aristophanes himself used the parisyllabic forms that epigraphic evidence suggests were common in 5th- and early 4th-century Attic: see, e.g., C.5 (a papyrus fragment possibly from the parabasis of the Heroes, as first suggested by Merkelbach 1967: see Kassel, Austin, PCG vol. 3,2, 178‒9; Pellegrino 2015, 201‒2), attesting the acc. plur. in -ως, already used by Aeschylus (C.2). The occurrence of acc. plur. ἥρως in classical Attic drama is, in fact, a possible reason for Phrynichus’ endorsement of this form; on the other hand, it is noteworthy that he rejects nom. plur. ἥρως (perhaps the expected regular form) despite its use by Aristophanes, revealing that his high regard for this author is not acritical (see Fiori 2022, 25‒6; Tribulato 2024, 78–9, observes that Phrynichus nonetheless finds ways to justify Aristophanes’ ‘slips’ even when he does disagree with his usage). Regarding the accusative singular, Philemon’s prescription of ἥρω (B.1) is in line with the Atticist preference for parisyllabic forms (with the exception of Phrynichus’ ἥρωες) and with the prevalent usage of canonical authors: apart from a single occurrence of ἥρωα in Plato (Lg. 738d.2), ἥρω is the usual form in Attic prose authors including Plato himself, Xenophon, and the orators. It should be noted that the ending -ω in this form may represent the outcome of the contractionContraction of athematic -ωα, but it also coincides with late-Attic 2nd declension forms such as τὸν νεώ (which, in turn, probably derive from the analogy of athematic nouns like ἕως, ἕω, according to Threatte 1996, 39). In general, some ancient grammarians deemed ‘apocopated’Apocope forms in -ω, such as ἈπόλλωἈπόλλων for Ἀπόλλωνα, typical of Attic (see entry ἱδρῶ, κυκεῶ, Ἀπόλλω, τυφῶ).

Schmid (Atticismus vol. 4, 583) observed that nouns like ἥρως, which oscillated between different inflectional patterns already in classic Attic, may still exhibit heteroclitic behaviour in Atticist writers. As an example, he cited the forms used by Philostratus, who ‒ for what concerns the direct cases ‒ oscillates in the acc. sing. between ἥρωα (3x) and ἥρω (7x) and between ἥρωας (2x) and ἥρως (2x) in the acc. plur., while he uses ἥρωες once as the nom. plur. (Atticismus vol. 4, 21). Other authors, however, appear to have avoided the parisyllabic forms, despite their old Attic pedigree, in favour of the athematic ones, which had the advantage of greater transparency: the tradition of Lucian’s text, for instance, only transmits acc. sing. ἥρωα (3x), nom. plur. ἥρωες (1x), and acc. plur. ἥρωας (4x). Meanwhile, the acc. plur. ἥρως is found in the Lucianic Encomium of Demosthenes (C.7), usually regarded as spurious.

E. Byzantine and Modern Greek commentary

Neither the ōs-stems nor the Attic second declension survived as such in Post-classical Greek. In the medieval and early modern periods, forms of ἥρως occur with various inflectional endings, including, for instance, acc. sing. ἥρων, gen. sing. ἡρώου and nom. plur. ἥρωες as well as a variant nom. sing. ἑρώας with a vocalism influenced by Italian eroe (see Kriaras, LME s.v. ήρωας). The classical nom. sing. ἥρως is also occasionally retained in higher-register texts, similarly to other vestigial 3rd-declension forms (see CGMEMG vol. 2, 342). The Modern Greek form ήρωας ‘hero’ has shifted to the α-stem declension as most other former 3rd-declension nouns, with a regular acc. sing. τον ήρωα and nom.-acc. plur. οι/τους ήρωες.

F. Commentary on individual texts and occurrences

(1)    Philemo (Vindob.) 394.20‒3 (B.1)

The first part of this entry from the Atticist lexicon of Philemon, transmitted in the extremely abridged recension of cod. V (Vindob. phil. gr. 172Vindob. phil. gr. 172, 15th century CE), was edited by Reitzenstein to restore two iambic trimeters (see entry Philemon, Περὶ Ἀττικῆς ἀντιλογίας τῆς ἐν ταῖς λέξεσιν on the traces of the original metrical structure). More importantly, Reitzenstein expunged the second part of the entry, which contrasts the above statement regarding the nominative plural forms with Phrynichus' doctrine, taken from the Eclogue (A.1). Indeed, this must be a later addition to Philemon’s entry: Reitzenstein (1897, 389) attributes it to an epitomator operating in the early Byzantine age, but it may also be due to the compiler of cod. V, a grammatical miscellany that at ff. 167^r–187^v contains a series of lexical entries taken from different sources, including the pseudo-Herodianic Περὶ ἡμαρτημένων λέξεων and Philetaerus, Philemon, and Phrynichus’ Eclogue, all presented anonymously and combined into a single work (see Fischer 1974, 25‒6).

(2)    Thom.Mag. 169.6–9 (B.4)

This entry in Thomas Magister's early 14th-century Atticist lexicon appears to draw, as it often happens, on Philemon for the prescription about the accusative singular forms; the following section, concerned with the accusative plural and nominative plural forms, closely follows Phrynichus’ Eclogue (A.1) in both content and form. Since Phrynichus’ prescription on the nominative plural was also added to Philemon’s lemma (see F.1), we may hypothesise that the copy of Philemon used by Thomas already contained this addition. However, given that the wording in Thomas is slightly closer to the text of the Eclogue than what is found in cod. V of Philemon, Thomas may have independently integrated this part directly from Phrynichus, on whom he draws heavily in the composition of his lexicon (although the omission of the name Φρύνιχος in Thomas’ version may signify early compilation). The slight difference may also be due to the fact that the compiler of cod. V based himself on a different codd. family of the Eclogue from that used by Thomas (who based himself on the q branch; see entry Thomas Magister, ’Ονομάτων Ἀττικῶν ἐκλογή).