7. Renewing Homer’s Licence: Accent and Phrasal Integrity1
The accentual practices in some of the earlier papyri appear to be arbitrary by comparison with the East Roman manuscripts. The grave sign seems to have been the weapon of choice for the correctors who wrote accent marks into the papyri. It properly says ‘suppress the rise here.’ It was a prohibition, not a prescription. But in a number of instances, the grave appears to mark all the syllables where there is not an acute, so that the syllable of high pitch is known by being the odd man out. But the acute itself is not marked!2 This is the interpretation of e.g., ὰιὸλὼπον for (standard) αἰολωπόν.3
It must be said, however, that this example is taken from a number which, together, are used to make the case that final oxytones were sometimes accented acute in mid-sentence in the ‘pre-Byzantine’ period. I find this possibility unlikely if the rule for completing the contonation within word boundaries was still in play. This rule depends on the fact that the contonation was ‘multi-moric’ and usually disyllabic, as well as restricted by the quantity of the ultima. More likely is that the successive graves were warning students against accenting alternatives, in the knowledge that the final syllable would be suppressed automatically. Highly likely is the possibility that Greek orators and composers worked to control the release of accents for rhetorical emphasis, using pauses and enclitics, sometimes producing long stretches within lines and sentences with no full contonations. I have pointed by way of example to the opening period of Sophocles’ ‘Ode to Man’: πολλὰ τὰ δεινὰ κοὐδὲν ἀνθρώπου … 4
In the papyri there is generally no teachable descriptive method apparent, as one finds in grammar books and in practice in Venetus A, but an ad hoc case-by-case and line-by-line approach; a seemingly unsystematic scattering of accentual assertions often after longer stretches of letters carrying no prosodic instruction at all. This is quite unlike the consistent word-level, propositional approach, based on Alexandrian scholarly authorities, of the Venetus A.
The use of the acute accent among the correctors responsible for accenting the papyri also needs to be approached afresh in light of the new theory. The acute does not originally and properly mean ‘here is the accented syllable,’ in the sense of the most dynamically prominent syllable in a word (or phrase for that matter). It only means ‘here the pitch rises’. Neither does it necessarily mean ‘here is the highest pitch in a phrase’, despite Nagy’s pleading, on the basis of certain late, explicitly melodic settings. Properly, in descriptive terms, the acute sign means ‘the contonation begins here.’ The locally most prominent syllable is often not the one where the high pitch is reached, but the one following: hence in reference to texts of the classical era, the acute can be as much a herald, as it were, of the word-level accentual event, as the repeated grave signs in some extant papyri preceding an unmarked acute.
One case where Nagy’s analysis is hampered by the late, monosyllabic, accented/unaccented approach, is with his take on ΚΡΑΤΎϹΑΡΓΕΙΦΌΝΤΗϹ (κρατύς Ἀργεϊφόντης) from the Bankes Papyrus (Iliad XXIV.345). He sees the acute on κρατύς, ‘strong’, as more evidence that in ‘pre-Byzantine’ Greek it was possible for final oxytones to be pronounced in mid-sentence, and not suppressed (‘graved’). What might the acute mean in light of the new theory? By itself, an acute could signify that a pause was felt here; despite the close attachment to the noun-phrase compound following, ‘Slayer of Argus’, it is not impossible, though unlikely, that this singing corrector wished to offset the cultic invocation of Hermes (Ἀργειφόντης) which fills the line from the diaeresis to the close. But if there was no pause, and the svarita or down-glide was still felt, the released acute requires a down-glide on the first syllable ᾽Αρ- crossing the word-boundary, causing dynamic prominence on that syllable. This may also be the case if the grave involved a slight rise in pitch.
On this understanding it may be that the oxytone preceding a polysyllable was sometimes ‘activated’ so as to produce a dynamically prominent initial syllable on the word following. In addition, when elision occurs, perhaps this indicates a merging of words (‘prosodic units’) such that boundaries evanesce. Consider a case like the phrase πατέρ’ ἐσθλόν (Odyssey 2.46, in context πατέρ’ ἐσθλὸν) where intuitively the writing convention, showing acute on the penult despite elision of the final, suggests that the contonation was completed on the initial syllable of ἐσθλόν. This initial would thereby be the prominent syllable (barytone) in the phrase. Where elision of the ultima unambiguously occurs—that is, the alternative reading of prodelision seems impossible or ill-advised, as here—it seems the only interpretation that allows πατέρ’ to be so accented is the one that sees ἐσθλόν as its ‘enclitic’ completion. There may also have been a phrasal, word-like integrity in such a Homeric collocation as κρατύς Ἀργεϊφόντης that encouraged such a marking and pronunciation within an extended word-boundary. Ἀργεϊφόντης itself seems to be a ‘frozen’ collocation of words that had acquired a new internal accentuation.
Equally possible, however, is a reflection in this case—and other cases in the papyri of mid-line or mid-sentence oxytone marking—of the fact that the contonation was no longer operant, and the accent had become monosyllabic. The contonation ordinarily was not allowed to cross word boundaries, because the down-glide needed to be completed within the word. If the historical accent had become monosyllabic, this restriction would no longer apply. In point of fact, the turning grave of ultimate oxytone syllables in mid-sentence would be as sure a sign as any that the original contonation and its word-boundary rule was still operant.
Recall the observation Nagy himself makes, that ‘already by the time of Aristarchus, whose floruit was the middle of the second century BCE, unaccented vowels were shortened while accented vowels were lengthened.’5 He cites evidence that the change to a stress accent was already under way in the time of the Alexandrian scholars. It is therefore quite unclear why he invests such authority in the accentual corrector of the Bankes Papyrus, who apparently worked in Elephantine some hundreds of years after this ‘sea change.’ One should have thought that far the simplest explanation for the peculiarities of his implementation in this papyrus—as, for example, his whimsical use of circumflexes on the antepenult discussed earlier—as well as this acute on κρατύς where the contonation would require a grave—is that the accent had become monosyllabic, whether circumflex or acute, and therefore proximity to word boundaries had no relevance to their performance. Since the text of the Venetus A preserves the rules of the recessive contonation, where the word ‘barytone’ in the usage of the scholiasts still has its common sense meaning, we should understand that the ethos of preservation in its scholarly pedigree, under the threat of a living and permanent language change, extends back to its primary authorities in Alexandria, who may already have been dealing with the changed prosodic state of Greek in their daily lives.
It is manifestly the case that the East Roman text of the Venetus A has preserved a prosody that was no longer current for contemporary Greek in Constantinople, including Greek versifying. When its chosen marks for the vowel mora of high pitch are understood to mark the onset of a contonation, I have demonstrated that the prosodic shapes so produced, marriages of pitch change and quantity in their permutations, for the first time make coherent harmonic sense of the lines of hexameter poets.6 As has long been apparent, this does not happen when one simply stresses the accent marks, or sings them on a higher pitch than the neighbouring unmarked syllables: in these implementations no discernible syncopation or reinforcement of the meter results.
I do not see how best to take the Bankes Papyrus as a clue to anything but the pronunciation and peculiar melody-making styles of its own age, or even merely its own editor. All the same, the papyri should be examined for each use of acute on the ultima in the context of an induced barytone at the beginning of the following word—in the new parlance—in case there is evidence there that the contonation of the ancient prosody sometimes worked in this way, inducing secondary prominence at the beginning of polysyllables—a practice not directly indicated in the Venetus A or modern texts.
The ‘phrasal integrity’ tentatively hypothesised here to help explain a pronunciation κρατύς Ἀργεϊφόντης, showing acute rather than grave, is not a remnant of some archaic integrity, sloping slipperily into the dreamscape of ‘traditional formulas’. Fans of ‘formulaic phrases’ as windows to the lost music of an oral tradition, or the melodic licks and clichés of an improvised oral composition, do not find what they need in Homer. What may be imagined here is the extension of the ‘word’ as a prosodic unit backward, as it already is extended forward by enclitics, so that, as is the case forward with enclitics, double contonations are possible in the rearward direction as well.
Here is Allen’s fifth rule for dynamic prominence in Greek:
A preceding element separated from the prominent element is also (secondarily) prominent.7
He found that in addition to the required terminal stress, syllables like the second in Ἀργεϊφόντης, separated from the prominent ultima by an ‘element’ (either ⏝⏝ or ⏤ ), are also prominent. There is no Alexandrian marking for this secondary accent. According to the descriptive law induced by Allen, secondary accent separate from the terminal accent is a feature of Greek words based on his study of syllable placement in relation to poetic ictus. Allen’s study arose from the patterns of syllabic reinforcement of the ictus at the codas of lines of stichic verse. Again, such doubling of word-level prosody is not marked rearward in polysyllables under the marking system used in the East Roman texts and in modern editions; only forward with enclitics.
One notes, however, this observation of William Wyatt, following one by A. Tsopanakis:
His observation is acute, and can be illustrated by the phrase ἡμετέρῳ ἐνὶ οἴκῳ ἐν Ἄργει (Iliad I.30), in which the final ῳ before the preposition is once shortened, once maintained as long.8
This -ῳ- exhibits the same extralinguistic pressure we see in Ἆρες Ἄρες, long in the thesis and short in the arsis. This is enough of an account for those who are satisfied that Homer’s music has its origin in dance. But it leads Wyatt to make a distinction between internal and external sandhi:
In this phrase, a formulaic phrase one imagines, the rules of external sandhi, which call for shortening of the final vowel in hiatus, do not apply with the first occurrence of the preposition … Formulas are in fact words, and therefore obey the rules of internal sandhi. Or, put more neutrally, in formulas and with certain words the rules of external sandhi are not obeyed or need not be obeyed. In the case in point, instead of a five-word phrase as printed, we have a three-word phrase: ἡμετέρῳἐνὶοἴκῳ ἐν Ἄργει, i.e. ἡμετέρῳ+ἐνὶ+οἴκῳ ἐν Ἄργει.9
Wyatt’s notion that ‘formulas are in fact words’ is not directly in reference to accentuation, but it serves as a way of acknowledging, from the perspective of word-level prosody, that the rules for such prosody interdepend on any rules for the determination of ‘words’ and their boundaries (ultimas). In the lemma discussed earlier, τὴν δὲ βαρυστενάχων, βαρυστενάχων is the only portion that can be claimed from the manuscript to be in scriptio continua, and it is equally possible as not that it registers as a compound word—or what Wyatt here describes as a ‘formula’ with internal sandhi. In this extended envelope of a ‘phrasal word’, properly a ‘phrasal prosodic unit’, we perhaps accept a greater likelihood not for the compositional reality of formulas, but for a more than theoretical reality for Allen’s secondary accent.
Plato in the Cratylus, composing in the age of scriptio continua, hypothesises the formation of nouns (ὀνόματα) out of clauses (ῥήματα)—e.g. Δίφιλος > Διὶ φίλος, ἄνθρωπος > ἀναθρῶν ἃ ὄπωπεν, 399a-c—in a way that fully presupposes the notion of words as integral prosodic units; this is long before their representation as such in manuscripts like the Venetus A. The process is one where a new word with a new accentuation is derived from a phrase where the original accents are responsive to the original word boundaries. Original letters are removed in the process, and some syllables change from pronunciation with a rising pitch to a down-glide, while others suffer the opposite change. Students who labour at their augments and endings in order to parse Ancient Greek sentences, ought to take heart that words and word-level prosodic melody were real things for the composers of texts, before their presentation in East Roman manuscripts as separate graphic phenomena.
Socrates describes the required changes of accent from the phrase to the derived word as a way, it would seem, to make his etymologies more plausible.10 If, therefore, in the lemma discussed earlier, βαρυστενάχων (or βαρυστεναχῶν) is to be understood as a word formed from a phrase (βαρὺ στενάχων, στεναχῶν), it should not be unexpected if a papyrus of Homer turned up showing a double contonation: βαρύστεναχῶν or βαρύστενάχων. The ‘fusing’ of the two words would close the ultima of βαρύ, allowing its contonation to be completed within the syllable. Allen’s stress rules assign prominence to the ultima of either version of the compound, and this is also true under the new theory, despite the change in the position of the final contonation, because the location of the svarita or down-glide in pitch remains in each case on the same long vowel. But the stress rules also assign ‘secondary prominence’ to the compound’s second syllable. No prediction is made, but no surprise would be registered, from the perspective of the new theory, at finding such a secondary marking in this location.
Perhaps in polysyllables—but also certain word-like collocations including name-and-epithet phrases, and other polysyllabic candidates sometimes described as ‘formulas’—there is possible a supension of the rules of sandhi, like correption, and secondary accentuation separated from the ultimate one. But it is unclear as yet what if any absolute rules there may be for the transformation from separate words to compounds: do the elements retain the original position of the onset of their contonation, or are such compounds re-conceived prosodically, as Socrates perhaps suggests in the Cratylus? Consider that if κρατὺς Ἀργειφόντης is such a collocation, the accentuations κρατύϲαργειφόντηϲ (as actually written in the Bankes Papyrus, but therefore actually prominent where bold), and also κρατυϲάργεϊϕόντηϲ, would not conform to Allen’s rule 5 for stress. This rule predicts prominence for the ultima—which is in line with the new theory of the accent—but secondary prominence on the segment γεϊ, which would be expressed under the new theory as γέϊ—which is not consistent with the written accentuation of the phrase in the Bankes Papyrus.
The producers of such manuscripts as the Venetus A, however, are methodical in focusing only on the terminal accentuation at word-level. In my own performance practice, I shall follow their judgement, in allowing the accentual melody of noun-and-epithet phrases to blossom purely from their constituent words and normal sandhi. But in some cases where the text unambiguously shows an acute on the penult where the ultima is elided, I shall obey the instruction: the contonation will be completed upon, and emphasise, the beginning of the word following. In general, it should be clear that elision of the ultima, where it must be posited in a text rather than prodelision, as well as shortening of the ultima, represent a definite breach of a word’s prosodic form, which in Greek depends on the quantity of the ultima.
Not all elisions find so straightforward a prosodic resolution, however, as the case of πατέρ’ ἐσθλόν. Sometimes they do indeed present phonetic difficulties. If we are to take the received instructions at face value, the use of elided particles, which release rising pitch accents, sometimes leave no room whatsoever for a down-glide, often because the ultima is elided. Consider ἡμεῖς δ’ οὔ νύ τι, or κάκ’ ἔρεξεν, or εἴτ’ ἄρ’ ὅ γ’. The effect heard in οὔ, κάκ’ or ἄρ’, with a second pitch rise immediately following, is not infrequent in Homer, perhaps a rapid-fire acute effect suited to the imitation of live speech, seemingly in defiance of the quasi-mechanical rule that what goes up must also come down within the word. What else can an honest performer do but obey direction? Homer, and his speakers, have a licence, and licence is the opposite of lawlessness.
Indeed, there is nothing to prevent our seeing a prosodic integrity in phrases as a creative Homeric characteristic, in sharp contrast with the view that sees him as a passive purveyor of ‘traditional phraseology’. The obverse of Plato the etymologist after the fact, is Homer the famous wordsmith, the ancient world’s greatest compounder and coiner of words. A lesson from Plato is that the joining of once separate words, erasing or eliding their former boundaries, is of necessity also a revising of their prosody that must release a latent, non-arbitrary musical potential.
It is saddening to think that generations of his students may have been robbed of the ancient world’s greatest wordsmith, the extraordinary namer and phrase-maker who championed Swift-Foot Achilles, Poly-Wily Odysseus and the Drunken Deep,11 to be left with a repository of traditional oral memory cards. It is past time to restore the fecundity of his musical conscience and consciousness, and to renew Homer’s licence. It is oral theory that has missed its expiry date.
Parts originally in A. P. David, ‘Singing Homer’s Spell: The Disyllabic Contonation and the Proposition Made by East Roman Manuscripts’, Dramaturgias: Revista do Laboratório de Dramaturgia da Universidade de Brasília, 19:7 (2022), 805-66. Note that this chapter represents both a retraction and replacement of the original section entitled ‘Secondary Accent’.
Jennifer Moore-Blunt, “Problems of Accentuation in Greek Papyri,” Quaderni Urbinati Di Cultura Classica, no. 29 (1978): 137–63.
Ibid., 141 [Ox.Pap.2369.ii.28].
see Sophocles’ Antigone, 332; discussion in A. P. David, The Dance of the Muses: Choral Theory and Ancient Greek Poetics, Oxford: Oxford University Press, 2006, 253.
Gregory Nagy, ‘Traces of an ancient system of reading Homeric verse in the Venetus A’, in Casey Dué (ed.), Recapturing a Homeric Legacy, Cambridge MA: Center for Hellenic Studies, 2009, 141.
see David, Dance of the Muses, esp. Ch. 4: ‘The Form of the Hexameter: The Origins of Caesura and Diaeresis’, 94-137.
W. Sidney Allen, Vox Graeca, 3rd edn., Cambridge: Cambridge University Press, 1987, 135-6.
William F. Wyatt, ‘Homeric Loss of /w/ and Vowels in Contact’, Glotta 72:1/4, 1994, 127.
Ibid.
David, Dance of the Muses, 86-7.
πόδας ὠκὺς Ἀχιλλεύς, πολύμητις Ὀδυσσεύς, ϝοἴνοπα πόντον.