Capotos ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 ISSN 2159-3159 Article copyright held by the author(s) and made available under the Creative Commons Attribution License CC-BY https://creativecommons.org/licenses/by/4.0/ Short Accusatives in Hesiod: A Diachronic Approach to an Un-Homeric Feature Spiridon-Iosif Capotos MONG THE LINGUISTIC FEATURES that seem to be almost exclusive to the Hesiodic text in the wider corpus of Greek hexametric poetry, the occurrence of a small but telling number of accusative plurals with a short vowel in the thematic declensions is probably the most significant and cer- tainly the most studied. There are six sure occurrences in the Theogony (60, 267, 401, 534, 653, 804) and three in the Works and Days (564 = 663, 675) in which the ending -ăs is used with stems belonging to the so-called first declension. This declension com- prehends in Greek two distinct groups of Indo-European nouns, namely those ending in a thematic -eh2 and the feminine nouns that are composed with the apophonic suffix ih2 / ̯ieh2.1 In historical Greek the ending -ᾰς for the accusative plural appears only in those dialects in which the phonological outcome of the original ending *-ns in sandhi before consonant (*-ăns + C > -ăs + C) has been extended to all other cases and become general, i.e. in North-Western Greek (NWG) and in the Aeolic dialects of the mainland (Thessalian and Boeotian) but not in Ionic, the closest relative of the epic dialect.2 1 The topic has an extensive bibliography; the most recent discussions can be found in A. C. Cassio, “Introduzione generale,” in Storia delle lingue letterarie greche (Milan 2016) 84–86. 2 The latest discussion of this isogloss is in A. C. Cassio, “The Language of Hesiod and the Corpus Hesiodeum,” in F. Montanari, et al. (eds.), Brill’s Com- A SPIRIDON-IOSIF CAPOTOS 377 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 If we begin our analysis with the assumption that ungram- matical forms in epic poetry are not simply instances of ‘poetic license’ but rather the products of analogical innovations result- ing from generations of oral composition, then we need also to assume that they depend on the formulaic system through which this poetry is composed. The analogical exceptions resulting from the adaptation of traditional formulas to different in- flectional contexts were later overgeneralized by the poets of following generations. This means that a kunstprachliche explana- tion of the short accusative plural requires us to determine what kind of context would have allowed the poet to develop a greater formulaic flexibility by integrating into his poetic grammar a metrically advantageous artificial form. I argue that we have evidence in Hesiod that the declension in the accusative of pre- existing and grammatically-sound formulas has produced these aberrant forms.3 1. Previous interpretations of the phenomenon Scholars have explained the presence of these short accusa- tives in the transmitted text as one of the following: a) under the influence of a continental poetic tradition in which Doric dialects played a more important role than in the Ionic tradition. b) the result of the direct influence of spoken dialects on the epic language, as consistent with the non-Ionic background of Hesiod or the singers of the Hesiodic tradition. c) the outcome of an internal analogical process which occurred within the artificial language of epic poetry at a certain stage of its development. panion to Hesiod (Leiden 2009) 179–201. The original coexistence of the two forms is shown in Cretan and Argolic, where the generalization has not been carried out and both the long and the short forms are preserved. 3 By formula I here mean “a group of words which is regularly employed under the same metrical conditions to express a given essential idea,” as first formulated by Parry: The Making of the Homeric Verse. The Collected Papers of Milman Parry (Oxford 1971 [1928]) 80. For discussion of formulaic modifica- tions see J. B. Hainsworth, The Flexibility of the Homeric Formula (Oxford 1968). 378 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 The first theory is as old as the first discussions of Hesiod’s language, and it is already present in the scholia: [1] Schol. Hes. Theog. 267: Ἀρπυίας Δωρικῶς συνέστειλε· τὰς γὰρ εἰς ας ληγούσας αἰτιατικὰς αὐτοὶ συστέλλουσι (Hesiod) Arpuiăs shortened in the Doric fashion: for the (Dorians) shorten the accusatives which end in -as. Accordingly, scholars have often interpreted these accusatives as Doricisms and attributed them either to the influence of spoken dialects of the NWG group or to a mainland poetic tradition. The great dialectal studies of the nineteenth century present this view,4 and Ahrens has also attempted to link Hesiod’s hexa- metric poetry to that of Delphic oracle. However, substantial evidence points in the opposite direction, namely that the oracular tradition depends on epic poetry, and it is now com- monly agreed that, if indeed there was an influence, the influ- ence came from the Homeric and Hesiodic poetry.5 A second group of studies has explained these forms as the result of an artificial process of ‘poetic assimilation’ in which the Boeotian bards tried to compete with contemporary lyric poetry by adopting some of its forms in their performances, among which there would also have been the accusatives in -ăs.6 Lyric and epic, however, are clearly distinguished ‘genres’ already in their first attestations and present their own clear linguistic con- ventions and Sitz im Leben. Most importantly, even if we were to 4 K. W. Goettling, Hesiodi Carmina (Henning 1843) 12; H. L. Ahrens, De Graecae linguae dialectis II (Göttingen 1843) 177–181; A. Rzach, “Der Dialekt des Hesiodos,” in Jahrb.f.class.Phil. Suppl. 8 (1876) 400–401. 5 See A. Morpurgo Davies, ‘‘Doric Features in the Language of Hesiod,” Glotta 42 (1964) 138–165; Cassio, in Brill’s Companion 179–201. A list of short accusatives in Delphic inscriptions can be found in Schwyzer, DGE 325. 6 C. Tsagalis, “Poet and Audience: From Homer to Hesiod,” in F. Montanari et al. (eds.), La poésie épique grecque: métamorphoses d’un genre littéraire (Vandoeuvres (2006) 79–130; A. Ercolani, Esiodo: Opere e Giorni (Rome 2010) 28–29, 349–350; L. Lulli, “Questioni aperte di una dizione epica: la mistione linguistica dell’epos di Esiodo,” SemRom 5 (2016) 195–215. SPIRIDON-IOSIF CAPOTOS 379 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 accept the possibility that lyric influenced Hesiod, what seems unconvincing in this theory is the extreme scarcity of these ungrammatically short accusatives, to the point that almost all occurrences, as we shall see, can be explained either by metrical necessity or formulaic modification. Finally, a third approach to the problem, to which the present work ultimately belongs, has rejected the traditional dialectal explanation and has seen these features as the result of an in- ternal development of the epic tradition. Scholars such as von Blumenthal, Davies, and Wyatt all agree in attributing the -ăs in our text to analogy, but vastly disagree in determining the con- text and causes of this analogical process.7 Von Blumenthal thinks that forms like Arpuiăs, aetăs, tropăs might have resulted from the influence of the athematic declension, where the ending -ăs is perfectly regular (*C-ns̥ > C-ăs). Moreover, he suggests that in lines like the two following, where thematic and a-stem ac- cusative plurals are closely juxtaposed, the quantity of the vowel -a- is irrelevant to the meter, and no palpable difference would have been perceived by the poet if both accusatives had been pronounced with a short vowel: [2] Theog. 935 δεινούς, οἵτ᾽ ἀνδρῶν πυκινὰς κλονέουσι φάλαγγας. ⏑ ⏑ - ⏑  ⏑ - ⏑ ⏑ - x Terrible, who drive tumultuously the dense (pykinās) ranks (phalangăs) of men. Op. 55 χαίρεις πῦρ κλέψας καὶ ἐµὰς φρένας ἠπεροπεύσας. ⏑ - ⏑ ⏑ - ⏑ ⏑ - x You feel joy now that you have stolen the fire and tricked my (emās) mind (phrenăs). Morpurgo Davies, instead, in her attempt to demonstrate that all Doric features in Hesiod are either archaisms or ascribed to formulaic modifications, considers short accusatives the result of 7 A. V. Blumenthal, “Beobachtungen zu griechischen Texten,” Hermes 77 (1942) 103–104; Morpurgo Davies, Glotta 42 (1964) 138–165; W. F. Wyatt, “Short Accusative Plurals in Greek,” TAPA 97 (1966) 617–643. 380 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 an alternation between forms that occurred before C or V. The original accusative plural ending (-ανς) had in fact lost the nasal element in preconsonantal position due to the simplification of consonant clusters in sandhi (ăns + C > ăs) and retained it instead before a vowel (ăns + V > ăns). In terms of relative chronology this phonological modification had to take place after the ninth century, as it postdates the Attic-Ionic change of the long central vowel ā into a middle front ē. Otherwise the second compen- satory lengthening would have caused the forms before vowel to become -ēs (ăns + V > ăns > ās > ēs).8 According to Morpurgo Davies, when some of the formulas ending with an accusative plural of the a-stem with a preconsonantal outcome found themselves in prevocalic position, especially before the caesura kata triton trochaion, the short ending was maintained, and these short accusative plurals were developed. Finally, Wyatt has proposed a three-phase process in which, because of the influence of a-stem masculine nouns such as ἰππότης, which also present a vocative ἰππότᾰ sometimes used as nominative, Hesiod had come to feel that a-stem declension could be characterized by a short vowel and he extended this feature also to the feminine nouns, creating an analogical -ăs for the a-stem declension. In addition to being uneconomical, these analogical interpre- tations remain open to criticism and do not explain why the process that created these artificially short forms should not also be attested in Homer, if we are indeed dealing with an internal and perhaps even Archaic development of the epic tradition. It seems likely that either geographical position or chronology must be considered, and a more comprehensive explanation still needs to be found. Von Blumenthal’s work problematically assumes that poets 8 It is important to point out that not all scholars agree on the absolute chronology of this Attic-Ionic modification, and some think that it is not a post-proto-Greek development. On the topic see J. Rau “Greek and Proto- Indo-European,” in E. J. Bakker (ed.), A Companion to the Ancient Greek Language (Boston 2010) 171–188. SPIRIDON-IOSIF CAPOTOS 381 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 who were even sensitive to overlength while composing (see 389 ff. below) could ignore the phonological distinction between long and short vowels which remained stable in all dialects for cen- turies. Wyatt’s argument is instead weakened by the fact that it is based on unprovable assumptions, namely that Hesiod re- interpreted the quantity of the characterizing vowel of the first declension because of a smaller group of masculine a-stem vocatives with -Ø ending. Moreover, there is only one a-stem masculine noun in Hesiod that presents a short accusative plural, µεταναιέτας in Theog. 401, out of the nine occurrences of the feature. And Morpurgo Davies’ theory suffers from the complete absence of any attestation in Hesiod of a short accusative plural within a formula before feminine caesura, and thus it is impos- sible to prove that the feature is a case of formulaic transposition. Another explanation must therefore be provided.9 2. Proposal: formulaic inflection and diachronic development In demonstrating that analogical developments of the diction are ultimately responsible for these artificial forms, we should first consider how formulaic inflections work in both Hesiod and Homer, with a few examples that are relevant to our case. Formulaic inflection can often happen without altering the grammar of the poetic language, as in [3] where the feminine participle ἔχουσαι is substituted for masculine ἔχοντες: [3] Hes. Theog. 75 ταῦτ᾽ ἄρα Μοῦσαι ἄειδον, Ὀλύµπια δώµατ᾽ ἔχουσαι ⏑ - x These are the things the Muses sing, holding (echousai part. pres. nom. plur. fem.) the Olympian chambers Il. 2.13 … Ὀλύµπια δώµατ᾽ ἔχοντες ἀθάνατοι ⏑ - x The immortals holding (part. pres. nom. plur. masc.) the Olympian chambers However, a formulaic modification can also create ungram- 9 Similar objections are raised by G. P. Edwards, The Language of Hesiod in its Traditional Context (Oxford 1971) 146–155. 382 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 matical scansions such as in the accusative gaieochon ( ˘ - ˘ ˘ ) in Theog. 15, in which the diphthong -ai- is scanned with the second element -i- treated as the onset of the following syllable. This phenomenon, which is normally interpreted as correptio, ulti- mately depends on the need to decline the formula.10 [4] Il. 13.43 ἀλλὰ Ποσειδάων (⏑ - - -) γαιήοχος (- - ⏑ ⏑), ἐννοσίγαιος But Poseidon (nom. sing.), earthmoving (gaiēochos nom. sing.), the Earth-shaker Hes. Theog. 1511 ἠδὲ Ποσειδάωνα (⏑ - - - ⏑) γαιήοχον (⏑ - ⏑ ⏑), ἐννοσίγαιον, And also, Poseidon (acc. sing.), earthmoving (gaiēochon nom. sing.), the Earth- shaker Even a possible variant γαήοχος motivated by the occasional loss of ι in diphthongs before vowel would be indirectly ‘caused’ by the need to inflect the formula. What concerns our study is that these modifications are the result of an attempt to fit a formula rather than a single word into the hexameter.12 It is important to clarify that this is not an exclusively Hesiodic process, and that the Homeric diction is rife with such examples. The presence of artificial forms in Homer due to formulaic inflection, however, has not caused scholars to suspect the in- fluence of different, unattested, spoken dialects or traditions. Moreover, the fact that ungrammatically long athematic 10 Correptio is rarer within the word than at word-end and, according to M. L. West, Greek Metre (Oxford 1982) 11, reflects an Ionic tendency of the language. This is a traditional example of Hesiod’s modification of formulaic prototypes. 11 This is the MSS. text and accepted by M. L. West, Hesiod: Theogony (Ox- ford 1966); other editors have emended to the metrically sound γεήοχον which however does not appear elsewhere in either Homer or Hesiod. So e.g. H. G. Evelyn-White, Hesiod, Homeric Hymns and Homerica (Cambridge [Mass.] 1914). 12 See C. D. Buck, Introduction to the Study of the Greek Dialects (Boston 1928) 29–30. Cf. P. Chantraine, Grammaire Homérique I (Paris 1948) 173, on the sim- ilar alternation between the usual βαθείης and the ‘shortened’ forms βαθέης (Il. 5.142, 21.213), βαθέην (Il. 16.766). SPIRIDON-IOSIF CAPOTOS 383 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 accusative plurals appear also in Homer because of formulaic inflection has received surprisingly little attention. The phrase νῆᾱς ἅλαδ’ [5] (cf. οἰκῆᾱς ἀλοχόν, Il. 6.366) provides us with an example of the same analogical process that we notice in Hesiod involving a-stems and consonant stems. Here the consonant stem ναῦς in the accusative plural has an irrationally long vowel (-ᾱς) which is metrically convenient and allows the poet to decline a formula originally in the genitive. [5] Il. 2.165 = 2.181 µηδὲ ἔα νῆᾱς ἅλα δ᾽ ἑλκέµεν ἀµφιελίσσας Do not allow them to drag their curved ships (neās athematic acc. plur.) into the sea Il. 14.100 σχήσουσιν πόλεµον νηῶν ἅλα δ᾽ ἑλκοµενάων (the Acheans) will not continue the war, once the ships (nēōn athematic gen. plur.) are drawn to the sea Rather than a metrical lengthening before caesura, this is a case of formulaic inflection combined with using the formulaic ἅλα δ᾽ ἑλκέµεν after an a-stem accusative plural, as both Chantraine and Hoekstra have argued.13 As a consequence, if we are looking to explain the presence of short accusatives in Hesiod, we need to examine those instances in which these forms are undoubtedly the result of formulaic in- flection. Particularly instructive in this regard is the formula µετὰ τροπὰς ἠελίοιο [6] which appears twice in the Works and Days and presents one Homeric parallel in the Odyssey, suggesting both that it is traditional, and, because of its single occurrence, that it is unlikely to be an immediate model for Hesiod. [6] Od. 15.403–404 νῆσός τις Συρίη κικλήσκεται, εἴ που ἀκούεις, Ὀρτυγίης καθύπερθεν, ὅθι τροπαὶ ἠελίοιο ⏑ ⏑ - ⏑ ⏑ -⏑ ⏑ - x 13 Chantraine, Grammaire 104; A. Hoekstra, Homeric Modifications of Formulaic Prototypes (Amsterdam 1965) 125. 384 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 There is a certain island called Syria, if ever you have heard (of it), beyond Ortygia, where are the turning points (tropai nom. plur.) of the sun Hes. Op. 564–565 εὖτ᾽ ἂν δ᾽ ἑξήκοντα µετὰ τροπὰς ἠελίοιο ⏑ ⏑ - ⏑ ⏑ - ⏑ ⏑ - x χειµέρι᾽ ἐκτελέσῃ Ζεὺς ἤµατα Whenever sixty wintry days, after the turning points (tropăs acc. plur.) of the sun Zeus has accomplished There is no need to presuppose that Hes. Op. 564 is more recent than Od. 15.404 but it is sufficient to point out that the formula in the accusative is a later analogical development based on the formula in the nominative plural. Moreover, this explanation would be ultimately coherent with Hesiod’s tendency to expand the flexibility of the inherited formulaic system.14 Another hint of this inflection is found in the ‘formulaic’ deinăs aētas of example [7] where again a single parallel with the Iliad provides a possible context for the formation of the short accusative: [7] Hes. Op. 675 καὶ χειµῶν᾽ ἐπιόντα Νότοιό τε δεινὰς ἀήτας and the upcoming storm and the terrible (deinăs acc. plur.) blast of the South Wind Il. 15.626 ἄχνῃ ὑπεκρύφθη, ἀνέµοιο δὲ δεινὸς ἀήτη (she) is hidden by the foam, and the blast of the terrible (deinŏs gen. sing.) wind Once again formulaic inflection is not in itself indicative of re- cent composition, as sometimes Homer presents artificial forms that Hesiod does not. This is the case for example of µέροπες (méropēs) ἄνθρωποι in Il. 18.288, which shows an artificially long vowel in the nominative ending as the result of the adaptation of the metrically sound formula µερόπων ἀνθρώπων (5⏑⏑-⏑⏑-⏑⏑-x), which is the normal form in both Homer and Hesiod (Hes. Op. 109, 143, 180). The only justification for such a form (meropēs) is 14 A more general tendency in Hesiod to decline inherited formulas in the accusative has already been observed by scholars; see the list of examples in Edwards, The Language of Hesiod 62–64. SPIRIDON-IOSIF CAPOTOS 385 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 the poet’s need to adapt a formula normally found in the genitive to the nominative. Even though the ending -ēs was never gen- eralized to the point of becoming an available metrical alterna- tive to the grammatical -ĕs, the process that produced this form is the same that we observe in Hesiod. It is likely that the existence of an accusative plural ending -ăs for the athematic substantives played a role in allowing the bards of the following generation to feel the short accusatives as grammatical enough to become a metrical alternative to -ās whenever the formulaic system required it. We still need to remember that the number of short accusatives in Hesiod is limited to a few occurrences. As expected in the context of a poetic Kunstprache the grammar often yields to the formulaic lexicon of the poet. When considered all together, these artificial adaptations are shown to be common to both Homer and Hesiod and to depend on analogy within a formulaic context that generally is shared by the two corpora. It seems to me that a conclusive explanation for the presence of short accusative plurals in Hesiod can now be attempted. These forms were the result of purely internal modifications of the epic language, which ultimately depended on formulaic de- clension in the accusative plural of preexisting formulas in the nominative. These forms were also probably influenced by the analogy with the short endings -ăs of the athematic declension, as has been partially suggested by von Blumenthal (n.7 above). We need to imagine an analogical proportion like the following: tropōn (a-stem): phrenōn (athematic) = tropăs: phrenăs This analogy presupposes that in the feminine genitive plural of the type tropōn the contraction between the -α- of the stem and the suffix of the genitive plural had already taken place (-άων > -ῶν), thus explaining why in the early uncontracted phase of the tradition such analogical process did not take place and occurred only at a later stage of the epic tradition. Overall, this theory has the merit of both providing an ana- logical explanation for the development of these forms in the poet’s mind (the poets of the Hesiodic tradition came to feel that 386 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 certain a-stem nouns could have an underlyingly short vowel in the accusative plural ending) and of pointing to the metrical and especially formulaic necessity for its employment in composition. This means that even occurrences like ἠυκόµους θ᾽ Ἁρπυίᾰς (Theog. 267), and µεταναιέτᾰς εἶναι (Theog. 401), which do not present a formulaic parallel either in Hesiod or in Homer, could still be explained as analogical formations. They are either the result of the overgeneralization of the artificial ending -ăs or of the direct inflection of a traditional formula which has not sur- vived into the extant epic poems, but which was still part of the formulaic system shared by Homer and Hesiod. In both cases it must be stressed that it is the formula and not the single word which was the original setting of these modifications. On a final note, it is worth discussing the only other occur- rence of an a-stem short accusative plural in the extant epic tradition outside of Hesiod: ἀθρόας (athroăs) in the Hymn to Hermes. [8] Hymn.Hom.Herm. 106 καὶ τὰς µὲν συνέλασσεν ἐς αὔλιον ἀθρόας οὔσας - ⏑ ⏑ - x and then he drove them in the cowhouse, being (ousas) in crowds (athroăs) As we would now expect, the ungrammatically short vowel in athroăs can be explained on the basis of formulaic modification involving nominative endings shortened in correptio epica in formulas such as ἀθρόοι ἦσαν (Od. 1.27) and ἀθρόοι εἶεν (Hom. Hom.Ap. 152). The testimony of the Hymn is all the more relevant because the short accusative is followed by an attic form ousas which is never attested in Homer and found only here and in Hom.Hom.Ap. 330. Earlier epic poems, including the ones of Hesiod, present only ἐοῦσα (< *h1s-ont-ih2) as the feminine present participle of εἰµί, and the juxtaposition of the two forms might vouch for the lateness of -ăs now that we have argued for its analogical development. In conclusion, even though it would be conceivable to imagine that at a certain point of the epic tradition, Hesiod or a group of bards operating in the mainland could have borrowed the mor- pheme -ăs from other contiguous Doric or NWG poetic tradi- SPIRIDON-IOSIF CAPOTOS 387 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 tions, both the use of this alleged borrowing in the context of preexisting formulas and especially the absence of analogous ac- cusative plural -ŏs of the thematic declension in either Hesiod or the Homeric Hymns make this alternative explanation unlikely.15 Only the analogy with the ‘naturally’ short accusative plurals of the third declension allowed the formation of these artificial forms, and the absence of ‘epic’ words with an -ŏs ending for the accusative plural made it impossible to decline formulas with thematic nouns in the accusative plural. Consider for example formulas of the type of the adjective ἀτάσθαλος (⏑-⏑⏑), such as ἀτάσθαλον ἄνδρα (⏑4⏑⏑5⏑) or ἀτάσθαλον ὀβριµοεργόν (⏑4⏑⏑5⏑⏑6x). Homer and Hesiod both present examples of inflection of these formulas (Od. 8.166 ἀτασθάλῳ ἀνδρὶ, Hes. Theog. 996 ἀτάσθαλος ὀβριµοεργός), and the position of the adjective before words beginning with a vowel also allows the inflection in the nominative plural with -oi in correptio epica in Od. 24.282 ἀτάσθαλοι ἄνδρες ἔχουσι (⏑4⏑⏑5⏑⏑6x), which can be compared to the genitive singular πατρὸς ἀτασθάλου αἴ κ’ ἐθέλητε (3⏑⏑4⏑⏑5⏑⏑6x) found in Hes. Theog. 164. Yet in Hesiod nothing like *ἀτάσθαλος ὀβριµοεργoύς or *ἀτάσθαλος ἄνδρας is ever attested or could be attested, even though it would be exactly the same formulaic context in which a-stem short accusatives developed. If the influence of a main- land tradition with Doric short accusatives happened to be the actual reason behind the presence of -ăs forms in Hesiod, then we would also expect -ŏs in similar metrical contexts. Yet, since in the Ionic tradition there was no way to analogically recreate short thematic accusative plurals, these forms are not found. Moreover, since oral poets are not grammarians, a new mor- pheme could have spread only through the borrowing of either 15 Unless one considers the Shield of Herakles Hesiodic and finds in the iso- lated evidence of lagŏs ([Hes.] Sc. 302) sufficient grounds to argue that thematic short accusatives were indeed part of the Hesiodic diction. Many linguistic features, however, as shown especially by Edwards, demonstrate that the Shield behaves differently from the rest of the Hesiodic poems, and it is likely to be a late imitation of Homeric poetry. 388 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 entire words or phrases from another tradition presenting this congenial metrical feature, and only then by analogy could have expanded to new contexts and words, yet these short accusatives are found only in words that are also Homeric and epic. It seems that, at least in this case, chronology and not geography provides a more likely solution to our problem. These short accusatives could then be added to the list of other morphological elements created by singers due to metrical necessity, such as distended aorist infinitives and datives in -eessi.16 Rather than attributing them to a geographical loca- tion, they should be considered characteristic of a later stage of the same Panhellenic hexametric tradition, a phase in which more artificial forms were generalized into the diction to allow greater formulaic flexibility.17 Since all these features are secon- dary developments and not particularly productive ones, it seems fitting to say that they are reflecting a more advanced, perhaps decadent stage of the diction.18 The conclusions that can be drawn so far are that: short accusatives in Hesiod are not dialectal but analogical forms. they were produced as a consequence of formulaic inflection; whenever a formula with a nominative plural in correptio epica was declined in the accusatives it maintained a short ending according to the rule of the epic Kunstprache and then was gen- eralized in other positions. the relatively unproductive use of a-stem short accusatives might vouch for the recent development of this feature. 3. The issue of overlength in Hesiod and the Hymns A final obstacle one needs to face in order to counter the 16 On the infinitives see A. Nikolaev, “The Aorist Infinitives in -έειν in Early Greek Hexameter Poetry,” JHS 133 (2013) 81–92; on the datives, A. Blanc, “Langue épique, parler des aèdes et datifs en -εσσι,” in F. Biville et al. (eds.), Autour de Michel Lejeune (Lyon 2009) 137–151. 17 See E. Passa, “L’Epica,” in Storia delle lingue letterarie greche 137–139. 18 See A. Hoekstra, The Sub-Epic Stage of the Formulaic Tradition: Studies in the Homeric Hymns to Apollo, to Aphrodite and to Demeter (Amsterdam 1969). SPIRIDON-IOSIF CAPOTOS 389 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 dialectal explanation for the short accusative plurals is the tendency of the Hesiodic poems, observed by Edwards, to allow a pattern -V̄C + #C consistently more often than the Iliad and the Odyssey, whenever accusative plural endings are involved.19 So far we have discussed short accusatives as a limited phenom- enon that affects only a few, mostly formulaic, phrases. But if this tendency of closing the syllable by position after a thematic ac- cusative plural is somehow related to the previous problem of the short accusative plurals, then the number of occurrences of these forms could actually be extensive. There are two possible explanations for this higher ratio of superheavy syllables. a) Hesiod’s poetry indeed scans long the vowel of the thematic accusative plural endings, and we observe a certain insen- sitivity to overlength which would differentiate the Hesiodic tradition from Homer. b) The high ratio of accusative plurals before consonant dem- onstrates that the vowel in the endings -ας/-ους was pro- nounced short and that the influence of NWG dialects is much more pervasive than was originally thought. Parry had already noticed that in the Homeric poems the predilected longum is often a preconsonantal closed syllable with a short vowel and that overlong syllables are used more frugally.20 Hesiod, instead, “places a disproportionally large number of them before a consonant.”21 This would seem, at first glance, to suggest that all a-stem and thematic accusative plural endings in Hesiod were perceived as characterized by a short vowel unless we find a different explanation for this compo- sitional difference. Before proposing a different interpretation of this tendency, it is worth pointing out that: 19 Edwards, The Language of Hesiod 155–165. It is this different treatment of the accusative plural endings that ultimately causes Edwards to opt for a dia- lectal explanation for the short thematic accusatives. 20 Parry, The Making of the Homeric Verse 93–94, 113–114. 21 Edwards, The Language of Hesiod 158. 390 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 a) The only abnormal occurrence of superheavy syllables in Hesiod is that of -ous before C in the Works and Days, which accounts for 71% of total occurrences of the ending. b) The occurrences of the accusative ending -ēn, especially in the Theogony, before C are comparable to those of the accusa- tive plural endings. One way to solve this problem is to look to the Hymns for further evidence that this phenomenon is either limited to Hesiod or perhaps relates to a later development of the epic tradition. Tables 1 and 2 seem to answer our question.22 Theogony Works and Days Hymn to Apollo +V +C _# +V +C _# +V +C _# (a) -ᾱς 7 35 12 2 15 11 4 6 10 (b) -ους 12 39 14 6 27 5 4 14 6 a+b 19 74 26 8 42 16 8 20 16 Hymn to Demeter Hymn to Hermes Hymn to Aphrod. +V +C _# +V +C _# +V +C _# (a) -ᾱς 2 1623 4 5 18 4 6 9 2 (b) -ους 2 3 224 1125 18 11 8 6 4 a+b 4 19 6 16 36 15 14 15 6 TABLE 1: Number of occurrences Theogony Works and Days Hymn to Apollo +V +C _# +V +C _# +V +C _# (a) -ᾱς 13 65 22 7 54 39 20 30 50 (b) -ους 18 60 22 16 71 13 17 58 25 a+b 16 62 22 12 64 24 18 46 36 Hymn to Demeter Hymn to Hermes Hymn to Aphrod. 22 I preserve the structure of Edwards’ tables in order to offer the reader results that are immediately comparable to his. I am particularly grateful to the Boston University MSSP consulting service, Daniel Cunha, Yingmai Chen, Jianing Yi, and Yuchen Huang, for testing the statistical significance of the data I collected. 23 Lines 462 and 465 are suspected to be interpolations. 24 ἵππους in Hymn.Hom.Dem. 388 is often considered an interpolation. 25 In line 492 there could be an observed digamma: νοµοὺς (ϝ)Ἑκαέργε. SPIRIDON-IOSIF CAPOTOS 391 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 +V +C _# +V +C _# +V +C _# (a) -ᾱς 9 73 18 18 67 15 35 53 12 (b) -ους 29 43 28 27 45 27 44 33 22 a+b 14 65 21 24 54 22 40 43 17 TABLE 2: Percentage of occurrence With the sole exception of the Hymn to Aphrodite, the Hymns seem to present a treatment of overlength similar to what we see in Hesiod, allowing a cluster of consonants in sandhi after thematic accusative plurals consistently more often than in Homer.26 The topic would require more space, but it can be briefly suggested here that the different ratios of superheavy syllables shown by the Hymn to Aphrodite might point in the direction of an earlier composition, as argued by Janko,27 who, however, dates the hymn within Hesiod’s lifetime.28 That both the Hesiodic poems and the Hymns share a certain insensitivity toward overlength seems to suggest that the accusative plurals were characterized by a long vowel, as we would expect, and that they are a characteristic of a later stage of composition and, perhaps, an Ionic stage. Interesting in this respect is the fact that most superheavy syllables of type (b) in the Hymn to Apollo occur in the so-called Delian half of the poem, the one suspected to have been sewn onto an independent Pythian hymn by the rhapsode Cynaethus of Chios and therefore reflecting a more advanced and Ionic facies.29 I thus believe that Nagy was ulti- 26 Particularly striking are the ratios presented by the Hymns to Demeter and Hermes, where the occurrence of superheavy syllables in sandhi is almost identical to that of Hesiod. 27 R. Janko, Homer, Hesiod and the Hymns: Diachronic Development in Epic Diction (Cambridge [Mass.] 1982) 151–180. 28 Although the presence of possible references to Hesiod’s poems have recently led S. D. Olson, The Homeric Hymn to Aphrodite and Related Texts (Berlin 2012) 10–15, to question Janko’s chronology. 29 This theory depends on a Pindaric scholium and has been defended by many scholars, such as Janko, Homer, Hesiod 112–114; A. Aloni, L’aedo e i tiranni: Ricerche sull’Inno omerico ad Apollo (Rome 1989) 107–131; and M. L. 392 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 mately right in connecting this phenomenon with the fact that “Hesiod generally reveals fewer constraints, and hence less archaism, than that of Homer” and shows therefore “simply a greater tolerance for this type of overlength.”30 Given the above- mentioned evidence provided by the Homeric Hymns, this lack of inherited constraints appears however to have been a more general feature of a later stage of epic poetry to which Hesiod’s poems belong. There is another potential proof, drawing on the studies of Hoenigswald on the treatment of overlength in the Rigveda in comparison to Homer, that this tendency is the result of chrono- logical variation.31 Hoenigswald has successfully demonstrated that words such as kṣétra- “field,” hotrá- “libation,” mandra- “lovely” in these hymns present far more occurrences outside cadence than in cadence and that this reflects a way to cope with the “deeply uncongenial phenomenon of overlength” (562). This tendency to avoid superheavy syllables in clausula was observed in Homer by Parry and could provide a better insight into the different behavior that is shown in Hesiod and the Homeric Hymns.32 If we are dealing with an inherited PIE trait which reflects the state of the spoken language then we could be West, Homeric Hymns. Homeric Apocrypha. Lives of Homer (Cambridge [Mass.] 2003) 9–12. 30 G. Nagy, Greek Mythology and Poetics (Ithaca 1990) 61. 31 H. M. Hoenigswald, “Overlong Syllables in Ṛgvedic Cadences,” JAOS 109 (1989) 559–563, argues that in Vedic Sanskrit, “synchronic processes and diachronic changes keep overlength in check” (559) and that at least in the early period the distribution of words with overlength “hints at a vastly larger quotient of occurrences outside cadence over those in cadence in the aggregate” (562). The difference with the Greek situation could be explained by the fact that the hexameter presents a much more rigid structure than the Vedic trishtubh and gayatri which are both characterized by a free opening and a more fixed clausula. 32 Parry, The Making of the Homeric Verse 93–94. The first to observe the phenomenon of overlength was J. A. J. Drewitt, “Some Differences between Speech-scansion and Narrative-scansion in Homeric Verse,” CQ 2 (1908) 100–102. SPIRIDON-IOSIF CAPOTOS 393 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 observing in Homer and Hesiod the progressive neglect of the inherited avoidance of overlength.33 In fact, Cassio has recently argued that in Homer this in- herited tendency was lost considerably early, during the Ionian phase of the tradition, as proved by verses such as the follow- ing:34 [9] Od. 8.113 Ποντεύς τε Πρωρεύς τε, Θόων Ἀναβησίνεώς τε (Then rose) Ponteus, and Proreus, Thoon, and Anabesineus The synizesis after quantitative metathesis in Anabēsineōs (-āo > ēo > eō) presents a very Ionic case of overlength and shows how in later recitations overlength was not perceived as problematic in clausula. Therefore, it is not unlikely that the different quo- tients of occurrences of superheavy syllables in Hesiod and the Homeric Hymns might reflect a mutated phonetic taste. This different treatment of superheavy syllables was already begin- ning to appear in Homer, where forms such as πηκτὸν (pēkton) ἄροτρον replaced πᾰκτὸν (păkton) ἄροτρον in the later recitations and would end with almost complete insensitivity to overlength in the poetry of the Classical period. Finally, it is worth noting that the treatment of overlength can vary significantly based on the position of the syllable in the hex- ameter. Wernicke’s Law, for example, shows a clear hierarchy of syllable length that positions overlong syllables at the top of the preference in the fourth contracted biceps, and overlength in 33 See A. Byrd, The Indo-European Syllable (Boston 2015) 192–193. Further evidence that a similar development of phonetic taste took place in the Indian tradition can be found in the phenomenon of Vedic pluti. It consists in the protraction of already long vowels especially in the context of interrogative sentence. Pluta vowels had therefore three moras, just like overlong syllables, and were extremely rare in the Rigveda (only three cases), more common in the Atharvaveda, and significantly more frequent in the later Brāhmaṇa literature. For an extensive discussion of this phenomenon see K. Strunk, Typische Merkmale von Fragesätzen und die altindische ‘Pluti’ (Munich 1983). 34 Cassio, in Storia delle lingue letterarie 65–66. 394 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 all contracted biceps is fairly common, as West argues.35 There- fore, a better understanding of the treatment of such overlong syllables in the hexameter would also require a broader revalua- tion of overlength in Homer, in order to take into account the different treatment of superheavy syllables in different positions and avoid a dubious generalization of the phenomenon.36 In a recent study, Ryan has noted that ultraheavy and superheavy syllables are treated differently in the Rigveda and that the scale of metrical weight is a complex one.37 As a consequence, over- length is a more complicated phenomenon than is normally be- lieved, since not all syllables were equally overlong and equally avoided, and more work needs to be done on the subject to understand in greater detail the relation between Homer, Hesiod, and the Homeric Hymns. The short accusative plurals of the a-stem declension and the phenomenon of overlength are therefore likely to be unrelated traits, which both characterize the Hesiodic diction and that of the Homeric Hymns. We have then few reasons to suspect that most a-stem and thematic accusative plurals in Hesiod were pronounced with a short vowel. Short accusatives and over- length are linked only insofar as they are both suspected of being representative of a later stage of the epic tradition. Addressing the issue raised by Edwards has therefore allowed us to find new potential evidence for a diachronic interpretation of the un- 35 Greek Metre 39. See A. C. Cassio, “Overlong Syllables in the Epic ‘Adonius’, ” in F. Gallo (ed.), Omero, Quaestiones Disputatae (Milan 2016) 31–42, at 32 and 40–41. 36 According to Cassio, in Omero 40–41, Parry was also misguided by a simplified approach to the problem of syllabic overlength. 37 K. M. Ryan, “The Development of Diphthongs in Vedic Sanskrit,” JAOS 141 (2021) 289–296; see also “Gradient Syllable Weight and Weight Universal in Quantitative Metrics,” Phonology 28 (2011) 413–454, where Ryan argues more generally that in languages in which the weight is said to be exclusively binary the “sensitivity to additional weight contrasts” (413) often causes the poets of each tradition to compose according to multiple quanti- tative contrasts not just two, or three if we include overlength. SPIRIDON-IOSIF CAPOTOS 395 ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 Homeric features in Hesiod’s language. Conclusions “If we assume that Hesiod’s autobiographical remarks are rhetorical rather than true in a literal sense, we need not pin- point him in the direct vicinity of Mt. Helicon.”38 This paper has argued that even the most noticeable linguistic difference between the Hesiodic and the Homeric Kunstsprache can be explained as an internal development of the same epic tradition. It is now possible to state clearly that ‘Hesiod’ is not ‘choosing’ certain forms any more than he is choosing a formula. Instead, the Hesiodic tradition is continuously and consistently innovating a formulaic and thematic system of which it is neither an imitator nor, so to speak, a competitor but rather a true repre- sentative. Even in producing forms that are un-Homeric, the poets of this tradition seem to be composing as Homeric poets insofar as they are trying to adapt and inflect formulas that are shared by both traditions and overgeneralize tendencies that are already present in Homer. This diachronic approach to Hesiod’s short accusatives is also consistent with the other late, and Ionic, traits which are char- acteristics of his diction: thus genitives in -εω, -εων with metathe- sis and synizesis (< -αο, -αων), the neglect of word-initial voiced labial-velar approximant (digamma), and ‘short’ datives in -oις, -αις. All these features seem to reflect a stage of the epic tradition in which Western Ionic dialects, such as Euboean and Attic, and the Aeolic dialect of the mainland, Boeotian, had acquired a more relevant role.39 Even though the idea of a mainland dialect 38 H. Koning, “The Hesiodic Question,” in A. C. Loney et al. (eds.), The Oxford Handbook of Hesiod (Oxford 2018) 17–30, at 24. 39 See M. L. West, “The Rise of the Greek Epic,” JHS 108 (1988) 151– 172; A. C. Cassio, “Lo sviluppo dell’epica greca e il mondo euboico,” in M. Bats et al. (eds.), Euboica. L’Eubea e la presenza euboica in Calcidica e in Occidente (Naples 1998) 11–22. 396 SHORT ACCUSATIVES IN HESIOD ————— Greek, Roman, and Byzantine Studies 64 (2024) 376–396 continuum has been recently refuted,40 nevertheless these areas could have exercised a sort of collective influence on the last stages of the Ionic epic tradition, and close cultural contacts between Euboea and Boeotia are indeed well documented.41 Finally, this view of the Hesiodic poems is also consistent with the ‘biographical details’ given in the Works and Days, in which Hesiod’s journey from Boeotia to Euboea could perhaps be seen as an instance of self-representation of the epic tradition.42 The seemingly secondary presence of short accusatives in Hesiod has proved to have many potential implications for the understand- ing of epic poetry as a whole.43 March, 2024 Dept. of Classical Studies Boston University scapotos@bu.edu 40 See N. Pantelidis, “Boeotian and its Neighbors: A Central Helladic Dia- lect Continuum?” in G. K. Giannakis et al. (eds.), Ancient Greek Dialects from Central Greece to the Black Sea (Leiden 2017) 167–188. 41 See A. D. Debiasi, Esiodo e l’Occidente (Rome 2008) 17–38. 42 An extensive discussion of the rhetorical and traditional self-reflexive elements of Hesiod’s biography is in G. Nagy, “Hesiod and the Ancient Biographical Tradition,” in Brill’s Companion 271–311. 43 I am very grateful especially to Prof. Alexander Nikolaev, and to Prof. Jeremy Rau and Prof. Gregory Nagy for their important comments and critiques on the first drafts of this paper, a first version of which was presented at the Harvard Indo-European workshop. I am also thankful to Dr. Zachary Rothstein-Dowden and Prof. Brandon Jones for their invaluable help and to James Aglio for the truly fruitful discussions we had on this topic. A special thanks to Prof. Sadovski for his help on the Vedic parallels, and to the anon- ymous reviewers of GRBS for greatly valuable comments.