5. The Disyllabic Contonation and the Visible Word1
The new understanding of the relationship between lemma and proposition, as defining that between scholium and text in Venetus A, allows us to see the framed text in that document as an inevitable advancement in the demonstration of Homeric prosody. The division between lines and the separation of prosodic units (principally reflecting a terminal word-level prosody) with a descriptive economy in the use of accent marks to indicate the mora of rising pitch and its occasional suppression, has not so far been improved upon. Perhaps the Roman-era apparatus was a touch more transparent and user-friendly than that of the Oxford Classical Text.
Contrast this view with one taken by Gregory Nagy. About the lemmas:
… [w]hat was being quoted in upper-case letters from Homer originated from pre-Byzantine traditions of actually reading Homeric verse out loud, and so the use of this upper-case lettering system reflects an older and therefore more accurate way of reading the text of Homeric verse.
And about the scholia attached to the lemmas, together with the framed text itself:
… [w]hat was being written in lower-case letters as the text of Homeric verse or as the text of commentaries about Homeric verse originated from contemporary Byzantine traditions of rewriting what had been written earlier in the pre-Byzantine period, and so the use of this lower-case lettering system can be seen as a newer and therefore less accurate way of reading the text of Homeric verse or even the text of commentaries about Homeric verse.2
Nagy sees as ‘less accurate’ a text that displays what is known to be a word-level prosody for Greek, as in fact a word-level prosody—separated words, each with their own accents. He also reminds us that ‘newer’ also means ‘later’ to a classicist, and hence further removed from text produced closer in time to a source or sources vaguely older. These do not include, in his view, an original score of Homer’s music; rather, a style of written presentation of a late script that more closely preserved a product or products of a once oral composition and orally dispersed transmission—all of which was ‘pre-Byzantine’. He goes on to play his hand:
From the standpoint of my own argumentation … the Byzantine rewriting of pre-Byzantine texts was less accurate. It stripped away two kinds of information embedded in the lettering practices stemming from the pre-Byzantine traditions:
(A) The new writing practice of using a space for marking where each word was separated from the next word undid the old writing practice of scriptio continua, which had served to protect the integrity of the phrase and integrity of the intonation embedded in the phrase.
(B) The new writing practice of consistently marking the accents of words on a word-by-word basis undid the old writing practice of selectively marking the intonation of phrases on a phrase-by-phrase basis.3
Let us be clear that under (A) Nagy supplies no evidence to support the following concepts: ‘integrity of the phrase’ and ‘integrity of the intonation embedded in the phrase.’ The latter would seem to offer the only clue, in fixed melodies known to be associated with certain phrases, to any possible reality of the former; but no such fixed (or affixed) melodies exist. Allen’s descriptive rules for stress may in fact allow for a limited application toward certain phrases as prosodic units—we shall discuss further evidence in a future chapter—but the work done by the word ‘integrity’ here is sheer mystification. Obviously there are no phrases separated from other phrases in scriptio continua. It is quite absurd therefore to point to a script of continuous letters without breaks, augmented by sporadic accent marks, and see protected therein the ‘integrity of the phrase.’
Under (B) I also find that the ‘new writing practice’ did undo something about the ‘old writing practice’, but what was undone was precisely the old practice’s notorious inconsistency. Intonation in Greek is properly embedded in a sub-unit of ‘the phrase’, which is to say the word. This is true of the recessive accent characteristic not only of Greek but of Latin and Classical Sanskrit. (Vedic does not show this quality of being recessive, under rule, from the ultima of the word.) It is therefore word accents that make music out of the epic ictus, by alternately disagreeing and then agreeing with and reinforcing it in a recognisable pattern, as I have demonstrated elsewhere.4
Nagy’s thesis, or rather lemma without proof, is that the selective accent-marking in pre-Byzantine texts, extant in papyri but also in the lemmas to Venetus A, preserves hints of a melodic contour that signals the highest pitch peak in the performance of a phrase; for him this points to the integrity of such phrases in transmission. Back of this for him, naturally, is the twentieth-century Homerist’s magical unicorn, the traditional formulaic phrase as a unit of composition in Homer. Aside from any philosophical, poetic or anthropological dimension, Nagy’s dispute with what he calls the Byzantine manuscripts of Homer lands materially on a dispute about the prosodic unit of Homeric composition. He can provide no evidence, however, for any sort of fixed, separable melody attached to any sort of verbal formula in Homer. In the face of this, the Venetus A makes an implicit but ubiquitous case for the compositional units that divide themselves and separate themselves when you simply write them out: prosodic words. It is precisely words that show the most literal kind of prosodic integrity in Greek.
Central to his argument (in his view) is the claim that the Venetus A text of Homer, in showing accent marks for each separated word, reflects the development of Greek from a language with a pitch accent to one with a stress accent:
These differences between old and new writing practices in the pre-Byzantine and Byzantine periods respectively were caused by what may best be described as a sea change in the evolution of the Greek language. We can see the beginnings of this change around the second century BCE. Already then, an old system of pitch-accentuation was changing into a new system of stress-accentuation. Where the syllable once had a pitch accent, there was now a stress accent. (This is not to say, however, that the new stress-accent was not simultaneously a pitch-accent as well.) And the new system of stress-accentuation persisted in the Byzantine Greek language and even in the Modern Greek language of today.5
Nagy goes on to cite his own summary, 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’—a consequence of the new stress implementation. Nagy nowhere gives evidence, however, that this supposed ‘sea change’ in any way affected Aristarchus’ scholarly decisions about the Homeric text, upon which the ‘Byzantine’ Venetus A depends. Nagy clarifies what he means, so there can be no doubt about his own claim:
In the old pre-Byzantine system, the pitch-accentuation of words had operated within a larger framework, which was the intonation of the phrase that framed the words. In the new Byzantine system, by contrast, the stress-accentuation of words was now operating within the smaller framework of the word itself. 6 [Emphasis added]
It is clear from the last that the word-level accent displayed in the innovative East Roman Venetus A text is being claimed to be a means of indicating the stressed syllable in each word.
‘Where the syllable once had a pitch accent, there was now a stress accent.’ Where indeed? Which syllable does Nagy suppose once had a ‘pitch accent’? Was it characterised by a rising, level, or falling pitch? Is he in turn imagining a stress feature that distinguished between the moras of a single vowel?
The rank incoherence of this is immediately illustrated by the use of the circumflex, in Venetus A and elsewhere. What could it mean in a stress-accent system? Syllables are either stressed or unstressed; why distinguish between circumflex and acute, most especially on long vowels—for example, βουλή and Ἀχαιῶν? Stress is on the ultima in both cases; what’s the difference? Nagy is handicapped: he is evidently unaware of (or ignores) Allen’s historical description of the contonation, applied from Vedic to Homeric and classical Greek, which was monosyllabic in the case of the circumflex but usually disyllabic. The later Greek stress accent, by contrast, is always monosyllabic.
To be sure, tonal prominence also picks out only one syllable, but it is a relative characteristic; its manner of emphasis does not tyrannise over other syllables, reducing their vowel grade, intensity and duration, in the way that a stress feature does. Non-prominent syllables in ancient Greek also show changing pitch in the non-prominent part of the contonation, whereas changing pitch, duration and intensity, in descending order of importance as perceptual cues,7 tend to be monopolised by a syllable with a stress accent. Unstressed syllables tend in various ways to deteriorate, whereas syllables not involved in the contonation at all in ancient Greek, still maintain their vowel values and duration.
Consider the corrector’s marking of this line (Iliad XXIV.408) from the Bankes Papyrus, juxtaposed with a modern version based on manuscripts like the Venetus A:
ῆετιπαρνῆεςςινεμοςπαιςῆέμινήδη
ἢ ἔτι πὰρ νήεσσιν ἐμὸς πάϊς, ἦέ μιν ἤδη
Amy Koenig notes about the Bankes text: ‘… here three instances of η are marked with a circumflex, indicating a falling pitch after the rising one. The second of these is grammatically inadmissible on the antepenult.’8 This inadmissibility is of course true also under the new rule: if the down-glide is completed on the first syllable of νῆεσσιν, more than one mora follows the completion of the contonation.
Under the new theory, this is a violation. It is the accent system exhibited in manuscripts like the Venetus A that supplied the database for both Allen’s induced rules for dynamic stress in Greek, and his comparison with Vedic that resulted in the formulation of the pitch-accent rule via the contonation. It is the combination of these separate results from Allen that forms the basis of my new theory of tonal stress or dynamic pitch for ancient Greek.9 The combination of the contonation with different quantitative environments produced two distinct prosodic shapes at the level of the word: oxytone and barytone. There are three shapes to distinguish if one separates the circumflexes (perispomena) from other barytones. Each of them is dynamic in the sense that, like stress accents, they can reinforce and syncopate metrical beats. But they are both combinations of lexical pitch changes and lexical quantity relations. And unlike with stress accents, only the circumflex is monosyllabic. Oxytones and barytones are by nature disyllabic: oxytones require a short syllable or pause following the pitch rise; barytones are post-acute.
I think it fair to say that the corrector of this Bankes text was also, like Nagy, not familiar with the contonation, or the rules of accentuation exhibited by the later Venetus A text. He did not appreciate the significance of the circumflex. An acute in the antepenult position would not violate the rules; it is therefore possible that there was no phonetic difference for this corrector, and his era, between the instruction from a circumflex and that from an acute. This is also the case under the later stress-accent: both signs mean ‘accented’. This line may therefore be of significance for the raw history of Homeric settings, or the development of Greek prosody, but for a modern prosodic editor or would-be performer it reflects a methodical inconsistency that renders the Bankes Papyrus, and its compiler, a dubious source. Nagy would have it that ‘the accentual markings made by the ancient diorthōtēs or ‘corrector’ of the Bankes Papyrus show that he was truly a master of correct poetic pronunciation.’10 But upon inspection, it is not excluded in the case of this line, that the corrector’s child had at it with the Egyptian equivalent of a crayon.
It is unclear why differences between the corrector’s use of accent marks for singing students, centuries later, and the accentuation of Homer seemingly quoted from Alexandria as lemmas, might be thought to preserve a melodic contour that was itself antique in relation to Aristophanes of Byzantium. It might be worth remembering that the Bankes Papyrus was produced when the New Music of Euripides was already ancient history. The practice of setting historical texts to new music was one which drew attention to a perceived disconnection between the new melodies and the original word-level prosody. I wrote:
… a lyric melos was made up of three things: speech, harmony, and rhythm (Plato, Republic 398d). But after its composition, a lyric was preserved only as speech, as a sequence of words (or strictly, letters) in a written text. In this form it could be quoted and interpreted … [A]ccording to the linguistic profile of Greek, to preserve a sequence of words is also to preserve a certain accentual harmony and a quantitative rhythm … [B]efore Euripides’ innovations, this harmony and rhythm were the originals, the constituents of the μέλος. As Plato says, the melic harmony and rhythm ‘follow’ (ἀκολουθεῖν) the melic word (λόγος) (Republic 398d); and in the Laws, the μέλος ‘suggests and awakens’ the rhythm (τοῦ δὲ μέλους ὑπομιμνήσκοντος καὶ ἐγείροντος τὸν ῥυθμόν, 637d) …
It was apparently fashionable in the revivals of the time of Plato’s Laws for the traditional melic texts, and possibly the fifth-century tragic choruses as well, to be treated by arrangers and performers in the new way, as if the words were music-less abstractions that could be set to a variety of ‘melismatic’ rhythms and melodies. This is why, when he wants to introduce some of the traditional poems and dances into the city (802a), the Athenian says his lawgiver must himself prescribe the harmonies and rhythms to which the μέλη will be set; for ‘it is a terrible thing to sing “off” with the whole harmony, or to “unrhythm” to the rhythm, having assigned unsuitable ones to each of the songs’ (δεινὸν γὰρ ὅλῃ γε ἁρμονίᾳ ἀπᾴδειν ἢ ῥυθμῷ ἀρῥυθμεῖν, μηδὲν προσήκοντα τούτων ἑκάστοις ἀποδιδόντα τοῖς μέλεσιν) (802e). The Athenian’s prescription is a rearguard action against what is often referred to as the New Music.11
On the evidence of the Bankes papyrus, its compiler and its prosodic corrector appear to be guilty of doing Plato’s ‘terrible thing,’ except in this case to the melody of Homeric epic. There are spellings that violate quantities, and of course there are those melismatic circumflexes. The practices here, apart from reflecting the critical loss of the disyllabic contonation in favour of a monosyllabic pitch or stress accent, seem more likely to reflect the desire to set old poems to new music—the fanciful circumflexes certainly are unprecedented—than a hidden clue to ancient Homeric melodies, dug up from the Egypt of the Common Era.
In its essence, the transition to the modern stress accent is easy enough to explain: the stress position is identical with the place where in the old system the voice rose in pitch. It is possible that the emergence of Greek as an international language after Alexander played a role in the transition, in that non-native speakers may have heard the rising high pitch at the beginning of the contonation as the accentual feature. Certainly if they got their books from Alexandria, this idea would have been reinforced: as today, one stresses the accented syllable. The down-glide would have lost its prominence, especially as the all-important reinforcement of quantitative ictus. In most cases it was not even indicated with a sign. We could therefore well imagine a stage where high pitch became the accentual signal, as well as the verbal cue for musical settings of what had been, in its manifestation under the Muses, a more percussive, orchestic phenomenon. (For a modern singer-songwriter, the guitar is equally a percussive and harmonic instrument.) This may have been a stage of the language that did not yet exhibit the concomitant features of stress, such as the shortening of unstressed syllables and weakening of vowels, in which the Bankes Papyrus perhaps best fits.
But in any case, the East Roman pitch accent system was still—all the same and as well—an indicator of melodic contour. Stress itself has a melodic contour; and the circumflex is precisely a graphic indication of melodic contour. Hence Nagy is asking the texts preserved in the Venetus A lemmas to point to a different, vanished melodic contour, somehow passed down in phrases, which was independent of the melodic contour manifestly inherent and indicated in the words—whatever the contemporary nature of the spoken accent, past or present. The hypothesis, if not denied reductio ad absurdum, is hopelessly fraught.
West points to a tradition of performing hexameter verse where there may well have been melody composed for hexameter verses that was not synchronised with the natural pitch pattern in the words:
The citharodic tradition that began in the seventh century and developed alongside the rhapsodic tradition did represent a radical departure. The citharode did not just spread the word accents over seven notes instead of [the] four [of the phorminx]: he employed real melodic structures in which the word accents were largely overridden …12
Nagy does not appear to be referring, however, to this ‘citharodic’ tradition when he invokes a ‘pre-Byzantine’ melodic contour for phrases in Homer. West interprets a musical fragment from Epidauros as showing a fixed melodic pattern repeating at the ends of a series of lines. But he distinguishes the practice of this citharodic tradition from that of the singing or declaiming of the rhapsodic tradition, which he links to practices he elsewhere argues belong to ‘eighth-century’ hexameter singing with a phorminx; he claims that
… the eighth-century aoidos sang epic poetry on four notes, the four notes to which his phorminx-strings were tuned, and that he followed the contours given by the word accents.13
With these views of West’s my argument sings in harmony: his eighth-century singer, like the later rhapsode, followed the word accents.
The pitch-accent distinction is that the acute on a long vowel indicates pitch rise on the second mora, circumflex on the first. But more importantly, the circumflex points to Sidney Allen’s solution for what Nagy calls pre-Byzantine Greek, that the Greek pitch accent was in all cases, not just circumflexes, a contonation involving a rise and a fall, as in Vedic. The Greek contonation, unlike the later high-pitch or stress accents, was usually disyllabic. Only in certain circumstances, most obviously in circumflected vowels, was it completed within a syllable. I repeat, the prominent syllable of falling pitch went unsigned in Greek except in the case of the circumflex; when it falls on a long quantity, I have called it the post-acute barytone. I maintain that this post-acute barytone, literally the heavy accent, was the Greek poet’s principal means of reinforcing the ictus brought to bear by the feet, at cadence points during a poetic movement—whether at the middle thesis (the third in the hexameter) or the end of a stichic line or, during the right- or leftward circling of a lyric strophic dance.
The East Roman manuscripts make a case, tacit but altogether vivid, for a new and true unit of speech and composition. In the beginning was the word, but this descriptive fact was not made articulable to the fluidly unifying consciousness of a speaker or composer—moving from sounds to syllables to phrases and sentences—until the habits of prosody were made to separate out these units of speech in visible writing. Modern assumptions and approaches to language, in teaching, learning and use, owe a debt to the Venetus A and the know-how that made it possible: part of this legacy is that the data have been preserved to allow insight into the music of Homeric μουσική. There is no call put out for formulaic phrases, or any hints of their phantom melodies, to register the dance of the composition and its effects: separated words, and their defining word-level prosody, will do.
Originally from 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; since revised.
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, 139.
Ibid., 140.
David, The Dance of the Muses: Choral Theory and Ancient Greek Poetics, Oxford: Oxford University Press, 2006, Chs. 4 & 8.
Nagy, 140.
Ibid., 141.
Larry Hyman, ‘Tone And/Or Accent’, in Donna Jo Napoli (ed.), Elements of Tone, Stress and Intonation, Washington: Georgetown University Press, 1978, 2.
Amy Koenig, “Homeric Accentuation: A Comparative Study of the Bankes Papyrus and Other Roman Papyri,” FirstDrafts@Classics, Center for Hellenic Studies, 2.
See David, The Dance of the Muses, esp. 71-2.
Nagy 143.
David, 32.
M. L. West, ‘The Singing of Hexameters: Evidence from Epidauros’, Zeitschrift für Papyrologie und Epigraphik, 63 (1986): 46.
Ibid., 45.