4. Digamma1
Modern classical scholarship is rightly proud of Richard Bentley’s inferred digamma. The presence of this phoneme despite its lack of graphic representation in the earliest extant texts or fragments of Homer is sometimes the only way to get Homer to scan—that is, to be metrical. Homer and Hesiod are already granted extraordinary licence with respect to quantities. I have pleaded that philology’s term ‘long by position’ (θέσει), describing vowels, should really be understood as ‘long in virtue of the thesis’ (θέσει) where short vowels in closed syllables can be performed as long when they occur on the long downbeat of the foot, but can be short elsewhere. There are cases (Ἆρες Ἄρες βροτολοιγέ, ⏤⏝⏝⏤⏝⏝⏤⏝) where no hidden digamma can aid the analysis: the alpha in Ares’ name is first long and then short purely because of its placement first in the thesis (downbeat) and then in the arsis (upbeat) of the same dactylic foot. But the inferred digamma generally saves Homer’s ghost from the charge of a wanton disregard for the natural quantities of Greek words when she deploys them in verse. And of course the existence of the digamma is thoroughly corroborated by comparative Indo-European. Where the digamma used to be in archaic Greek, Latin shows a ‘v’ which was pronounced in the classical era as a waw. The digamma is not always invisible: the υ component in classical Greek diphthongs, for example, sometimes reflects the one-time presence of digamma in the stem, as in the the masculine nominatives in -ευς.
This digamma needs to be reckoned with, when we consider the historical notion of ‘correcting’ the written text of Homer—whether in the context of scriptio continua or otherwise. The new theory requires that the realisation of the prosodic contonation, determining a most prominent syllable, depends in its rule on the quantities of the syllables following its onset (the rise in the pitch of the voice). Here again is a descriptive rule (under the new theory) for the recessive Greek accent:
The Greek pitch accent is a contonation, a rise followed by a fall in adjacent vowel moras. It can be monosyllabic or disyllabic. Where possible, pitch rise occurs on the antepenult; but no more than one mora may follow the syllable bearing the subsequent down-glide.
There is this stipulation:
The contonation must be completed within the word; when there is no following mora to bear the down-glide within the word boundary, extended by enclisis or punctuation, the rise in pitch is therefore suppressed (indicated in writing by turning the acute mark to grave).
The quantity of the syllable bearing the down-glide determines which part of the contonation, the sharp rise (ὀξύς, always over only one mora) or the heavy fall (βαρύς, sometimes over two) registers as most prominent. Hence the knowledge of quantities is essential for correct prosodic performance. This knowledge is not, however, directly recoverable from a written representation where ϵ can stand for short ε and long ει and η, or a score for Homer which does not include the digamma (ϝ).
Consider, for example, these entries from the front of the lexicon: ἀάατος (perhaps from *ἀ(ν)άϝατος*).2 The historical output is spelled with a slightly comic double hiatus; the clash of vowel on vowel without an intervening consonant (hiatus) was clearly something undesirable—perhaps forbidden—in Homeric music. What is more, the word is made to exhibit two different quantitative rhythms in the text of Homer: ⏝⏤⏤⏝ in the Iliad and ⏝⏤⏝⏝ in the Odyssey. The former is therefore made barytone on the penult; the latter oxytone on the antepenult. Does one simply assume that metre has forced the musical issue in each case? Or should the different quantities and pronunciations in the Iliad and the Odyssey versions indicate that the lexical entries should be separated and derived differently? In any case such internal hiatus cannot be ‘solved’ by orthography.
Next we have ἀᾱγής, also from the Odyssey, where the lengthened penult is said to indicate an initial digamma: ἀϝαγής. Here the digamma handily solves the hiatus, but it seems also to be asked somehow to lengthen the penult. More often, the digamma served to lengthen a syllable by straightforwardly closing it (e.g. ξεῖνος > ξένϝος). In the latter case the lengthened vowel (ε —> ει) compensates metrically so as not to affect the prosodic reinforcement. I claim that metrical lengthening in Homer only occurs before or with the onset of the contonation,3 as here. Perhaps there is less restriction on metrical shortening, even turning barytone into oxytone, as we see after the onset in ἀάᾰτος in the Odyssey.
This modern ‘discovery’ of the role of ϝ in Homer therefore rather problematises the correctness of the ancient prosodic correctors, pre-Alexandrian or otherwise. It does not occur, after all, in their corrected texts, but its absence is compensated for, metrically, by other means. To what extent, or for how long, might there have been a shared but unexpressed cultural osmosis between writer and reader and performer about the hidden presence and effects of the phoneme represented by digamma in Homeric prosody? There is no question that in the historical, post-Homeric period, Greek moved to lose certain intervocalic stops, notably -σ- and -ϝ-. This loss is reflected in the transmitted Homeric texts, but there is generally no vowel contraction because of the resulting hiatus, and the metrical properties of the remaining syllables are preserved. For all that Homer is famous for ‘metrical lengthening’, and shortening (Ἆρες Ἄρες), it is equally clear that the digamma closed a number of his syllables, and that he eschewed hiatus in mid-line.
The loss of digamma resulted in different phonemic outputs, together with various strategies then available to compensate for prosodic consequences in the different environments within and between words—all at some time prior to a Pisistratean recension of the written text. The linguistic chemistry involved in restoring those digammas, sometimes replaced in writing by movable nu and various particles, is not always a fixed science; there are alternate routes. And yet the modern inferred digamma, closing syllables and preventing hiatus, connects us to an original composition reflective of Homer’s native language and aesthetic, whose authenticity is best guaranteed by a conservatism demanded by an audience—first from rhapsodic performers in classical times, but then carefully observed by scholars in Alexandria. Indeed it would seem that the successful inference of the digamma, corroborated by comparative historical linguistics throughout the Homeric lexicon, would not have been possible unless the snapshot of that composition, which was preserved and edited in Alexandrian and East Roman texts, did not reflect an authentic and reasonably consistent translation into the digamma-free medium of historical Greek writing. An Urtext of Homer is as real as the digamma, but also as the genuity of ancient Homeric scholarship whose consistency and conservatism allowed its discovery through its various remnant effects. No smile without a Cheshire cat.
But what were these conservatives preserving? No modern edition of the Iliad or the Odyssey now prints the digamma where its presence would help explain Homer’s rhythm, or prevent hiatus, and where it is soundly corroborated by comparative historical analysis as once having been present. Nor has every case of apparent hiatus in the received text been ‘solved’ by the digamma or other means. It is assumed, however, that the earliest written texts reflect the contemporary state of dialectal pronunciation. So did Homer care or not care about the aural effect of hiatus? Did he or did he not fancy the -w- sound? The case of the Nikandre inscription shows movable -ν- being used to prevent hiatus where digamma had been lost. Martin West argues
[t]his shows that by that time [the mid-seventh-century] Ionian poets were already using movable nu to cure digamma hiatus. If one believes as I do, that the Iliad was composed and written down at about the same period, one will take this as adequate justification for leaving such nus in the text. It is not unreasonable to suppose that the insertion of particles such as γε, κε, τε, or ῥα for the same purpose began as early as the addition of the nu.4
Nagy critiques a perceived assumption on West’s part that the writing of the nu in this inscription bespeaks a practice reflective of the composition and writing of the Iliad itself. But Nagy does not address the need for what West calls a ‘cure’; does it after all make sense that the original composer was in the business of systematically curing a problem with his dialect, rather than composing to its strengths? Neither West nor Nagy, ‘multitext’ nor Urtext, addresses the question of greatest import, it would seem, for those keen to ground their theories in the exigencies and realities of performance: what was the state of the digamma in the practice, and indeed the consciousness, of Homer and his audience?
There are many cases where loss of digamma causes hiatus within a word, as in the name of the dawn, Ἠώς, or the examples given above from the beginning of the lexicon. Scriptio continua does not distinguish these internal syllables from external ones: avoidance of hiatus reflects an aesthetic at the level of the syllable, not the word. A performer does not have the luxury of remaining agnostic about the -w- sound. He has to choose, in saying Dawn’s name, between the scratch of a glottal stop (Eh-Os) and the smoothness of a glide (Eh-Wos). For an artist there is more than the difference of a movable -n- between a wanax and an anax.
The most straightforward assumption is that the digamma lived in the lingual or dialectal sound-world in which Homer composed his music. This is the common sense view for anyone undertaking to sing ἀάατος or Ἠώς. The ‘avoidance of hiatus’ is in fact the way grammarian philologists register a phenomenon that was actually caused by a cantor’s need, and desire, to articulate and voice the vowel of any sung syllable with a distinct, specific, initial consonant. It is not about separating vowels like naughty sheep, but articulating them as the bearers of melody in song. Against the sense of recent scholarship, the common sense must be that Homer composed with this hidden consonant opening (and sometimes closing) his syllables. I shall therefore deliver digammas where they need to be, on the assumption that they are, at the end of the day, where Homer wanted them to be because they sound better than does hiatus. This will be my choice for a ‘cure’.
When one considers the sources of the variants implied or explicit in the apparatus of the East Roman manuscripts—an Athenian vulgate, other city editions, classical fragments, generations of Alexandrian scholars—the situation of the Homeric text becomes comparable to that of the Shakespeare plays published in the mysterious literary Folio, alongside apparent working scripts prepared for actual performances a generation prior. The editors of a recent King Lear decided against producing a Shakespearean master text for students, and instead edited and published, in one volume, two separate plays with differing titles. But the role of the digamma takes Homer’s text to a metaphysical quandary deeper than the status of folio versus quarto; it takes the composition of Homer’s poetry to an era prior to that of the earliest dialects in which it was written down. All the same, a performer—whether bard, or citharode, or rhapsode, or modern student—has no use for metaphysical sounds, even if he is singing the hexameters of Parmenides. He must do more for the syllables of Homer’s music than merely infer ϝ. My own cure results, for what it’s worth, in neither a complete restoration of the waw sound (the fully metaphysical option) nor its elimination and substitution (the slavishly historical one), but in its rounding semi-presence—appearing and disappearing like Banquo’s ghost, or ens before aitches in English.
It is difficult to feel confidence with West that the Homeric poems were composed in an era when the composer had to devise ‘cures’ for the hiatus caused by some already historical loss of digamma. The avoidance of hiatus is rather an active aesthetic impulse, with a decided practical motivation for a singer. In some cases the apparent cures can be seen as interpretations of graphemes; for example, an original grouping ΞΕΝΟϹ could equally have been read ξενϝος as ξεινος, or ξηνος for that matter. But the movable nus and elided particles represent graphic interventions. They mean to compensate for missing sound. The consistency of such cures suggest system and method, and perhaps reflect a meeting of minds, as is mooted in a Pisistratean recension.
But it is impossible to know. This is in fact the mature position, not for experts or the faint of heart. What seems clear is that the hiatus-ridden, needlessly unmetrical text the Alexandrians later received was not, as some are keen to suggest, the dictation of any sort of original performance. The implication of the inferred, and absolutely necessary waw sound, is that in the first instance the text of Homer was a translation into a daughter language. The original composition, whether or not there was once a script or a recording, remains a multiplex and multi-dialectal mystery—and an aural wonder.
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.
Richard John Cunliffe, A Lexicon of the Homeric Dialect, Norman: University of Oklahoma Press, 1963, 1.
David, The Dance of the Muses: Choral Theory and Ancient Greek Poetics, Oxford: Oxford University Press, 2006, 155.
M. L. West, ‘The Textual Criticism and Editing of Homer’, in Glenn W. Most, ed., Editing Texts/Texte edieren, Aporemata II, Gottingen, 101; quoted by Nagy, Review of M. L. West, Homeri Ilias Vol. 1 & 2, Bibliotheca scriptorum Graecorum et Romanorum Teubneriana, Bryn Mawr Classical Review (2000.09.12)