1. Dictation Theories and Pre-Hellenic Literacy

1.1 Statement of the Problem

The enduring fascination of the Homeric poems attests to their undeniable artistic integrity. These are not haphazard products, cobbled together unredacted by one or more editors from independently preexisting songs. The architecture of their themes and plot construction evince sophisticated long-distance correspondences that I would not hesitate to label prolepses and analepses. The patent artistic integrity of the Iliad and the Odyssey must provide a fixed point to any theory of their genesis and eventual writing. On the other hand, we have ample, and in my view, incontrovertible, evidence that anchors them to the oral song culture of ancient Greece. These poems were composed in performance, with the aid of traditional themes and a sophisticated traditional diction, and without the use of writing. This inescapable conclusion follows from the internal evidence of the poetry, as it is illuminated by Milman Parry’s and Albert B. Lord’s studies of living Southslavic oral poetic traditions. [1] The attempts to chart a way between the Scylla of their primary orality and the Charybdis of their artistic integrity constitute the chronicle of twentieth-century Homeric scholarship. There are still those who hold to a substantially written composition. They believe in the historical reality of a poet ‘Homer,’ who, pen in hand, charted and executed the composition of the poems. But these are lone voices, unwilling to come to terms with the implications of Parry’s and Lord’s scholarship.
Of greater influence are those who accept the reality of a singer of exceptional skill—let us call him ‘Homer’—in complete command of the techniques of oral composition, who either himself wrote down or else dictated to an amanuensis an especially fulsome version of his poems. The differences of detail between versions of this theory are many: some think Homer [2] was illiterate and therefore could only have dictated to another; [3] others, following Lord, go further and assert that a working knowledge of literacy (as seems requisite for the task of committing almost thirty thousand hexameter lines to paper [4] ) would have denaturalized the poet’s oral technique and must be ruled out; [5] yet others point out that the skill of writing may be inorganically related to the techniques of oral composition, and that Homer’s writing down remains a possibility. [6] Common to all versions, however, is the notion that we should consider the resulting written texts as inorganically related to the oral compositional technique: writing will have affected, but not overruled, the singer’s traditional compositional modus operandi. From the point of view of Homeric poetry as a system, the writing down of the poems was not a diachronic development but a historical accident. [7] This book explores an alternative to such dictation (or autographic) theories, which I consider historically implausible and diachronically impossible. The author, and still the most eloquent exponent, of this alternative is Gregory Nagy, who in a series of books and articles has described what he calls “an evolutionary model” for the making of Homeric poetry. [8] Because I agree with Nagy’s proposal and follow it in its basic outlines, I devote this and the chapter that follows to a critique of the main dictation theories in print. I hope to make clear why I find dictation theories implausible in the extreme and Nagy’s alternative eminently preferable. Nagy combines the comparative evidence about the composition and performance of Homeric poetry with the evidence for its diffusion during the archaic and classical periods. Just as the dynamics of performance shaped the manner of composition (the creation of, and recourse to, traditional diction and themes), so also the dynamics of diffusion shaped the evolution of this poetic medium. The main comparanda are the living epic traditions of India. Fieldwork has shown that the greater their diffusion, the less occasional they become as the performer seeks the acceptance of a growing target audience. Wider diffusion is also correlated with the increasing professionalization of the performer. This fact is of obvious significance to the focus of this book. Choices of subjects and their particular variants, thematic emphases and de-emphases, are carefully tuned to the breadth of audience reception. One may readily appreciate the difficulty in mastering a repertoire of many detailed versions, each possessing the local color appropriate to a narrow audience. Also clear is the corresponding advantage of converging on a common denominator suitable for the most general kinds of audiences. As with all comparative work, the study of Indian epic traditions reveals both significant similarities and dissimilarities with the Greek ones. In particular, a semantic shift from hero to god is perceptible in those epic traditions that have experienced the widest diffusion and become most normative. This stands in contrast with the Greek focus on the hero’s mortality. But the insight remains valid that diffusion deeply affects the dynamics of reception that obtain between the performer and his audiences, and has the potential to shape his subject matter and composition.
Nagy also draws attention to the role that festivals may play in elevating local poetry to supralocal status. [9] This too is instructive, for it suggests that we consider the role of the Panathenaia in Athens and other relevant festivals elsewhere as centers of diffusion that may potentially attract supralocal performers and audiences. [10] One may envision the drawing of public and performers to a central venue as a centripetal force that subjects the reception of epic poetry to the common demands of the progressively integrated and standardized tastes of its festival audience. At the end of the festival, when outside performers leave and visitors return to their locations of origin, they act as centrifugal agents of diffusion, taking with them and advertising as they may those performances that have enjoyed the greatest success. When there are competing centers of diffusion, the process of homogenization to which recurring festivals give place creates over time recognizable local variants of supralocal projection, whose influence at regional and supraregional levels depends on the perceived prestige of the corresponding festival. In the ancient Greek case, I note the so-called ‘city editions’ of the Homeric poems (κατὰ πόλεις or πολιτικαί), which at first probably represented epic versions independent from the Athenian, but in the long run may have evolved into syntheses of locally received variants and the more canonical (i.e. Panhellenic) Athenian version. [11] The repeated convergence of performers and audience on a dominant venue matches the convergence over time of local variants on a comparatively more canonical, Panhellenic version. This pattern of diffusion is the key to more unified traditions. Because Athens in the sixth and fifth centuries BC (if not before) became the dominant center for the performance of Homeric poetry, [12] this book will focus on the performance practice of rhapsodes in Athens.

1.2 Albert Lord’s Dictation Theory

I start with Lord’s own dictation theory, first published in Lord 1953 and now conveniently reprinted in Lord 1991:38–48. [13] His towering reputation and centrality to the study of oral poetic traditions make his views the necessary point of reference of all later dictation theories. Familiarity with his formulation clarifies the critical ways in which he differs from later exponents of dictation, who have eagerly, but with doubtful legitimacy, used him to support their own theories. The central contention that issues from Lord’s work is that the hypothetical text that might emerge from the dictation to an amanuensis by an oral traditional singer remains a historical accident, organically unconnected to the workings of the oral tradition. Diachronically, it would be as trivial and inconsequential as it is historically implausible. This follows from Lord’s consideration of possible sources for the impetus to record the song in writing. [14] Even if we assume arguendo that it was available to them, accomplished oral poets would not think of turning to writing from their accustomed composition in performance. This is especially so when writing was still rudimentary. What would make them dissatisfied with the long-established mode of composition? Nor could they need writing as a mnemonic aid, since their training and life-long practice would have taught them to perform without the extraneous need to memorize a script. Nor can the impetus to memorialize a song provide a plausible motivation. When the tradition has been orally handed down to an oral poet and he brings it to life in performance, what would suggest to him the need for anything other than oral transmission to the succeeding generations? Why should he be anxious that his songs would be lost without a written record? Nor do oral traditional singers share our notion of ‘word’ (see below, §7.1), which would make them eager to preserve verbatim a particular performance.
Lord therefore concludes that the impetus must have come from an outside source, but he does not speculate about the historical circumstances of a possible dictation event or the ensuing transmission of the resulting transcript. He only notes that our texts may go back to “an early period of collecting” which is otherwise unknown. [15] Here, Lord writes presumably under the influence of his own field experience as a collector of living oral traditions. It is precisely his commendable refusal to speculate that is responsible for his historically implausible theory. If he had carefully considered the circumstances that must have obtained in archaic Greece, the technical obstacle to the alleged dictation, and the role that the hypothetical transcript would have had to play in the song culture to make its way through the archaic, classical, and Hellenistic periods into the medieval paradosis and, finally, into our hands, I am confident that he would have written otherwise. But we must not lose sight of the central implication that flows from his deep grasp of the workings of oral traditional composition in performance: so long as the oral tradition was vigorous and supple, no hypothetical record of one singer’s performance, whatever its quality, could have arrested the centuries-old rhapsodic practice and become a controlling canonical text that performers would have to memorize and perform by rote. [16] For Lord, dictation preserved the primacy of oral performance and guaranteed the culturally inorganic, marginal status of the resulting transcript—a transcript, in his view, whose existence was demanded by the surviving texts. Thus, his article on the oral dictated text was an attempt to harmonize the model of composition in performance derived from his fieldwork with the survival of the Iliad and the Odyssey as written texts. [17] The greatness of the man and his scholarship is not diminished by his failure to answer adequately questions that he never pondered. It is paradoxical that his authority has been borrowed to prop up dictation theories that are incompatible with his model of oral composition. Even if the implausible historical accident of dictation had actually happened, the resulting transcript would have been irrelevant to the diachronic development of the text of the poems and largely marginal to their transmission. [18]
Here it is appropriate to consider the impact of dictation on the practice of a singer accustomed to compose in performance before an audience. Those who do not dismiss Lord’s work out of hand, but still think that the Iliad and the Odyssey cannot owe their monumental dimensions to the ordinary circumstances of oral composition in performance, find an escape in what they think are the unhurried conditions of dictation. Once again, Lord himself seems to lend his aid: “It is impossible to believe that the Iliad and the Odyssey as we have them represent exactly the songs as actually sung in normal performance by Homer; their length and consequent richness of content, the perfection of their lines, suggest some reliance on writing”; and again: “The very length of the Homeric poems is the best proof that they are the products of the moment of dictation rather than that of singing. The leisureliness of their tempo, the fullness of their telling, are also indication of this method” (Lord 1991:45 and 46). But these comments might give the false impression that dictation comes easy to the bard, a welcome relief from the pressures of composition before an audience, a process that ensures the successful production of a high-quality composition. Nothing is further from the truth: “One might think that dictating gave the singer the leisure to plan the words and their placing in the line … . First of all, however, dictating is not a leisurely process; neither the singer nor the scribe has patience for long pauses for deliberation … . But more especially, a mood and a tempo … are established which produce the balanced utterances of the poet. Not conscious planning but the rhythm of that particular process of composition calls forth the structures” (Lord 1991:12). Lord’s corrective makes clear that, if dictation succeeds at all, it is not because it supplants the ordinary psychology of composition in performance by the conscious premeditation of the literate poet. Only the rhythm is changed, and, with the change of rhythm comes a tendency to ornamentation and structural elaboration. [19] But it is the exceptional singer who can successfully adjust to the unaccustomed rhythm, and the exceptional scribe who can move at the requisite speed and provide the stimulation and feedback the singer needs and would otherwise get from his regular audience.
For as long as recording devices were not available, many were the instances of failed attempts to record a dictated performance. As Lord (1991:12) notes in a marked understatement, “not all singers can dictate successfully.” To this, he adds the following caution: “It is vastly important that we do not make the unthinking mistake of believing that the process of dictation frees the singer to manipulate words in accordance with an entirely new system of poetics. Clearly he has time to plan his line in advance, but this is more of a hindrance than a help to the singer who is accustomed to rapid-fire association and composition” (Lord 1960:128). Given the ways in which dictation interferes with the regular practice of the singer, how it upsets the usual psychology and rhythm of composition, it is hardly surprising that the literature contains many examples of frustrating attempts to record dictated performances and their inferior products. Thus Radloff (1990:85–86) complains about the deficient compositions that resulted when the Kirgiz singers performed for dictation:
Unfortunately, I must concede that despite all my efforts I did not succeed in completely reproducing the singers’ songs. The repeated singing of the same song, the slow dictations, and my frequent interruptions slackened the singer’s excitement, which is often necessary for good singing. He could only dictate in a fatigued and lax manner what he had recited to me with fervor a short time before. Although I was generous with applause and gifts to encourage the singer, these could not make up for natural motivation. Therefore, the recorded verses have lost much of their freshness.
The great Serbian scholar Vuk Stefanović Karadžić met with comparable difficulties when he sought to record the songs of Old Milija:
[W]hen I met Milija, instead of happiness I found nothing but toil and trouble. Like nearly all the other ‘singers’ (who are nothing but singers), he couldn’t recite the ballads in the proper order but only sing them. And not only that; without spirits, he’d become so confused and decrepit … that he couldn’t even sing always in the proper order. When I saw this I could think of nothing better than to make sure that he sang each song to me several times until I’d got it well enough by heart to know when he was skipping something; then I’d ask him to sing it slowly to me, drawling out the words, and would write after him, as quickly as I could. When I’d copied down a song like this, he’d have to sing it to me again, and I’d look at my manuscript to see if it was all written down properly. [20]
It is important to quote at length from these field experiences, because the modern armchair scholar is prone to underestimate the difficulty of dictation and its chances of failure. These examples should make us reluctant to assume too readily that an alleged eighth- or seventh-century dictation might plausibly explain the genesis of the written Homeric poems. On this sole basis the historical accident of dictation cannot be positively ruled out, but the need of this scenario for the co-occurrence of a series of exceptional circumstances hardly recommends it as a plausible origin. Not only would we need an exceptional oral singer with incomparable mastery of his traditional practice of composition in performance; he will also have had the assistance of an exceptionally gifted scribe with the intelligence and skill to prompt his successful dictation; and he will have had to prove himself an extraordinary reciter of a prodigious poem under the abnormal and unprecedented circumstances of dictation.
Lord 1954:8 serves to emphasize how unfamiliar these must have felt to an oral traditional singer accustomed to compose in performance. Lacking the rhythm his instrument imparts to his verse and the critical acclaim of an audience to spur his effort, the singer finds dictation the most difficult of all methods of collection:
Without a good audience the singer tends to shorten his song … . The scribe must know how to induce the singer to form good verses, even if it means putting the gusle into the singer’s hands from time to time to restore the proper rhythm to the singer’s mind. Moreover, the scribe must also take the place of the audience … [H]e must assure the singer … that he is a discerning audience who will tolerate nothing less than the best which the singer can produce. This takes great skill; for the scribe is working against the evil of ennui both of himself and of the singer. Epic songs are long. In normal rapid performance they frequently take many hours. In the slow process of writing and of dictating line after line, hours are not infrequently stretched to days. The singer is unused to such slow composition. His mind frequently runs ahead several lines while the scribe is writing the last one which he dictated, and the scribe must be constantly alert to this propensity. Yet all this guiding of the singer by the scribe, this nursing of the song, must be done without the scribe in any way inserting himself or his ideas into the song. He may suggest that something isn’t consistent, but he can never advise what the correct way would be.
The hypothesis of an oral dictation of the Homeric poems in the archaic period requires the co-occurrence in a common setting of an extraordinarily skilled oral singer and an exceptionally sensitive scribe, together with the technology of writing and the means to defray the attendant costs. Extreme skepticism about the reality of such a remarkable historical accident seems justified. [21]
It remains to understand how the evolutionary model can account for the monumentality of the poems. One can understand the tendency towards comprehensiveness as part and parcel of the Panhellenic aesthetics of reception and its normative status. The intense cultural focus on the Homeric poetic tradition over against other poetry, which subjects the Homeric poems to constant elaboration and augmentation, is reflected by Lykourgos’ statement in Against Leōkratēs §102 that, of all poets’, Homer’s ἔπη alone should be performed at the penteteric Panathenaia. Consider in this light the way Karadžić describes the growth of heroic poems: “The poems have not immediately, in their beginning, become as they are, but one man begins and composes as he can, and then going from mouth to mouth the poem grows and becomes more beautiful, and sometimes shrivels and is spoilt; for as one man can talk more clearly and beautifully than another, so he can sing and speak poems.” [22] When shaped by convergent cultural forces of reception, long-lasting traditional recomposition in performance may foster the growth of a notional thematic totality, of which any performance only constitutes a partial view. A good example of one such convergent force would be the ‘Panathenaic rule,’ which would have required rhapsodes to perform sequential subject matter. [23] The suggestion of legal compulsion in these late testimonia need not be taken at face value. Diachronic developments in ancient Greece were often reimagined as the decisive action of one individual. It would not surprise us if what the rhapsodes felt as the growing insistence by the Panathenaic audience to perform Homeric subject matter in a particular sequence—the audience knows how to reward and how to punish—would have been attributed to the initiative of Solon or Hipparkhos. The eventual written text of such notional totality, what Nagy (1996b:76) calls “the singular marvel of ultimate [poetic] expansion,” would naturally expand to the monumental dimensions familiar to us. Aelian’s Varia Historia 13.14 makes clear that the ancients could readily conceive of the poems in terms of episodes. [24] What the prevailing taste might consider the highlights, at least, would probably be performed more often and receive a corresponding title. This would not lessen the notional totality which would at length result largely in the poems as we know them today. This diachronic development is the explanation of the poems’ monumentality which scholars, including Lord, have found puzzling apart from the historical accident of dictation. [25]

1.3 Richard Janko’s Dictation Theory

Richard Janko has elaborated his dictation theory most fully in Janko 1990 and Janko 1998a. He is primarily motivated by two convictions. The first, that in Janko 1982 he proved conclusively through statistical linguistic analysis that the Homeric poems were fixed in writing in the late eighth century BC. In a recent review, he confidently affirms that “this demonstration … has yet to be refuted,” [26] and this study has been very influential. Haslam 1997:80 serves to make the point. He quiets his misgivings about an early fixation of the Homeric poems with an appeal to the work of Janko. [27] The second conviction is that the Homeric poems exhibit the sorts of mistakes and inconsistencies that may be expected from the crucible of composition in performance. For an autographic text, these mistakes would be lapsus calami; for a dictated one, lapsus linguae.
But Janko 1982 is as influential as it is methodologically flawed. The matter is too complex to be reviewed here, and a detailed critique must be relegated to future work. But it is worth noting that he has not convinced West, who in the exposition of his own dictation theory [28] observes that from a demonstration that the language of the Iliad and the Odyssey is less modern than other hexameter poems one may not readily infer that they were composed earlier. One need not be wedded to a nineteenth-century view of a thoroughly artificial Homeric Kunstsprache to embrace the currency of ‘productive archaisms’ in a singer’s oral composition (cf. West 1995:204–205). [29] In his 2008 dissertation and a recent article, Brandtly Jones claims that the approach of Janko 1982 “is fundamentally and pervasively flawed” and that “his methods fail to date [the epic texts] in either a relative or an absolute manner.” [30]
Regarding his second conviction, Janko argues that a poem composed with the aid of writing would exhibit a “much smoother surface” (1998a:7) and he presents a few examples that illustrate the dictum nescit vox missa reverti (7–9). These he thinks sufficient to refute the performance-driven model of gradual textual fixation advanced by Nagy (cf. 1998a:12n63), who in turn has answered them in Nagy 2003:49–71. But it is worth commenting on an especially well-known case introduced in Janko 1990:328, 331–332 and elaborated in Janko 1992:99–100. [31] This concerns Ν 410–423, where Deiphobos strikes Hypsenor and ‘immediately loosed his knees’ (εἶθαρ δ’ ὑπὸ γούνατ’ ἔλυσε 412). This expression (with or without ὑπό, and with γυῖα or γούνατα) is used often for a fatal blow that dissipates the limbs’ vital force, e.g. at Δ 469 Ζ 27 Η 12 Λ 240 579 Ο 435 Π 312 400 Ρ 349 524 Ω 498 ξ 236 ω 381 (and presumably at Ε 176 Λ 260 Ν 360 Ο 291 581 Π 425 465). But sometimes it is clearly not (Π 805 Φ 406 425 Ψ 726), or at least by itself not necessarily, fatal, [32] even though parallel instances of the type of wound involved might point in this direction (Η 16 compared to Ε 43–47). The expression is further used simply to express weakness (θ 233), instability (σ 242), weariness (Η 6 υ 118), or feebleness induced by strong emotion (Ν 85 Σ 31 Φ 114 δ 703 ε 297 406 σ 212 341 χ 68 147 ψ 205 ω 345). Since death coincides with, or often follows, a fall in battle, ‘to loose the limbs’ becomes an easy metonymy for ‘to kill’. But Π 805, where the fall precedes the wound, and Φ 406 425, where death is an impossible outcome (not to mention Ψ 726, while wrestling), make clear that the stricter meaning (i.e. provoking the collapse of the limbs) can be evoked even in the context of fighting. Indeed, σ 238 242 explicitly play on the ambiguity of whether death attends the loosening of the limbs. [33] The parallels establish that it is not impossible, if perhaps surprising, to use the expression where a wounded fighter falls to the ground but does not die (at least immediately) from his wounds. That Deiphobos at Ν 416 and Idomeneus at Ν 447 should think so is understandable, but we need not take their perception or inference as normative. [34] In fact, Leaf (1900–1902:2.33) ad Ν 420–423 observes that it is unusual for two Greek heroes to leave the fight to carry a dead body back to the ships (they merely draw it within their lines to prevent despoiling); whereas this development is well motivated if they are carrying back a wounded leader. And he adds that while βαρέα στενάχων is otherwise used in the Iliad only of wounded warriors (Θ 334 Ν 538 Ξ 432), the Odyssey applies it to those in distress (δ 516 ε 420 κ 76 ψ 317), like the near identical βαρὺ στενάχων in the Iliad (Α 364 Δ 153 Ι 16 Π 20 Σ 70 78 323 Ψ 60). There is nothing in the context or diction of Janko’s alleged example of an irreversible performance mistake to force the interpretation that Hypsenor had to be dead and could not groan; or to suggest that in making Mekisteus and Alastor, his carriers, groan heavily the passage would necessarily be felt as departing from regular epic diction. It is impossible, in other words, to conclude that Aristarkhos’ στενάχοντε, his alternative to the vulgate στενάχοντα at Ν 423, must have been an emendation without textual support. [35] The scholiast’s judgment that reading στενάχοντα is laughable (γελοῖον), usually (and perhaps correctly) ascribed to Aristarkhos too, cannot decide the matter. The Alexandrian scholar did not understand that Homeric poetry had had its genesis through traditional recomposition in performance. Presented with the alternative, he (or his disciples) could judge στενάχοντα laughable on the basis of his own proclivity to integrate his disappointed expectation of a dead Hypsenor with the particulars of the context and his feel of Homeric diction. In creating either reading, a singer in performance would clothe his choice of a particular outcome (a wounded but living or a dead Hypsenor) with the language customary to fighting contexts according to his particular feel of the semantic boundaries of this conventional diction. One singer might feel unsuitable the temporary survival and groaning of a wounded hero whose limbs had been loosened; he might then produce στενάχοντε. [36] Another might not share the scruple and produce στενάχοντα instead. Varying degrees of tolerance for what are perceived as potentially problematic statements, and a corresponding willingness to accept (and harmonize them with the context) or reject them are typical of the reception of especially authoritative texts. One need only think of the charge, often levied by the skeptic, that the Bible is full of contradictions, an objection that many believers readily counter with an array of harmonizing strategies. In the case of Homeric poetry, the singer himself is an instrument of creation, transmission, and reception, and the production of multiforms in performance is subject to the same considerations of acceptable subject matter and diction. I might add that I find implausible the notion that ‘Homer nodded off’ in the short space of thirteen lines; or that (incredibly) having nodded off, his skilled scribe (for skilled he must have been to produce the alleged dictated texts) would have failed to bring this to his attention, as Nikola would do during his transcriptions of Southslavic songs. [37] Under the circumstances of dictation as envisioned by Janko, the notion of an ‘irreversible’ mistake (nescit … reverti) is psychologically implausible. It requires that the singer never hear again, or ask to have read back to him, what he has previously dictated. Otherwise, would he not have noticed the ‘laughable’ incongruity and instructed the scribe to make the necessary changes? How are we to explain his lack of interest in the product of what would have been, under any scenario, both a stunning departure from his regular practice and a prodigious feat of historical innovation? [38]
Janko (1990:330) asserts that the technical obstacles to the writing down of the Iliad and the Odyssey in the late eighth century have been “much exaggerated.” He notes that the alphabet had been derived from the Phoenicians no later than 775 BC (which precedes by about half a century his date for the Iliad); he declares that writing materials were available, with the sole support of Herodotos 5.58; [39] and he suggests that contact with the Near East, where written epics (he claims) circulated widely, might have prompted a similar feat among the Greeks. The first two matters are best taken up individually, and I will do so below. For now, I simply note, first, that the date of the derivation of the alphabet, while an obvious terminus post quem, is not the primary concern. Of greater interest is why the alphabet might have been created and to what uses it might have been put at such an early stage. [40] Second, that, even if Herodotos can be made to imply the currency and ready availability of leather for writing, it is very unlikely that the Greeks would have used this expensive substrate for the almost certainly commercial purposes for which the alphabet was invented. [41] This suggests that at the earliest stages of alphabetic adoption, users would have not resorted to the use of leather. But that is the same early stage at which the poems are supposed to have been written down. That hardly recommends leather for transcripts of epic performances, to say nothing of the substantial amounts of substrate required for the thousands of lines in the poems as we know them. But if we resort to the more reasonable conjecture of the use of papyri, plausibility is on the side of papyri not being available to the Greeks at this early stage. [42]

1.4 Written Epics from the Near East

Janko’s third argument, that written epics circulated widely in the Near East, needs further qualification. When the particular nature of this written record and the social circles of its diffusion are understood, this argument loses much of its force. [43] I can best make my point by restricting the argument to one of the NE epics that received the widest diffusion, the epic of Gilgamesh. [44] What follows is heavily indebted to George 2003, the most recent authoritative editor of the epic, and all page references are to this work unless otherwise stated. We must bear in mind, however, that the centralized administration typical of NE rulers, with its extensive archives and a professional class of scribes, marks an immediate and important departure from the setting of Dark-Age and archaic Greece. Actions and developments that are readily intelligible in a court setting that values written literacy highly cannot, without justification, be assumed to apply to Greek Dark-Age and early-archaic centers of power. Another factor that makes NE epics a parallel of doubtful applicability is the unlikelihood that the Greeks would have been directly exposed to Mesopotamian culture. NE literary and cultural practices mediated by the civilizations of the Levant would seem of higher relevance. A final fact of stark difference: even so long a poem as the standard version of Gilgamesh did not have more than about three thousand lines; [45] compare that to the almost 15,700 of our Iliad and over 12,100 of our Odyssey! The combination of a significantly smaller poem and a social class dedicated to writing and archiving makes the recording of Gilgamesh and its written transmission a diachronic development of NE culture, not (as would be the corresponding writing of Homeric poetry at the alleged time) a historical accident, impossible to rule out but of extreme improbability.
There are four primary stages in the composition and transmission of Gilgamesh: the Sumerian stage (down to 2300 BC), and the Old-Babylonian (OB) period (2000–1600 BC), the Middle Babylonian (MB) and Assyrian period (1600–1000 BC), and the Neo-Babylonian (NB) and Neo-Assyrian period (1000–400 BC). [46] To these, George adds consideration of Gilgamesh outside the cuneiform tradition. The oldest fragment of the epic comes from Nippur and dates to the Ur Ⅲ Neo-Sumerian period (2100–2000 BC). Much of the extant Sumerian literature at the time was set down in King Shulgi’s academies of Nippur and Ur and was adopted as the regular curriculum of OB scribal schools. [47] Most extant manuscripts of Gilgamesh go back to eighteenth-century OB scribal apprentices, but George is reasonably certain that the literature represented there goes back to the Ur Ⅲ period as court entertainment (7). By the OB period it was limited to scribal circles (8). The Babylonian (i.e. the Akkadian, not Sumerian) version of Gilgamesh was not part of the scribal curriculum in the eighteenth century. And yet, a few scraps of an Akkadian Gilgamesh among the mass of Sumerian school tablets show that it had already joined the written tradition. George conjectures that “people of this time could have been familiar with Gilgameš stories in the vernacular Akkadian from an oral tradition. The Gilgameš motifs found on terracotta plaques of the Old Babylonian period support such a view, for they are more likely to reflect people’s knowledge of orally transmitted stories than to witness popular familiarity with a written version” (17). He further believes that the poem was performed in Akkadian in court circles and the marketplace, and that it was widely known by all social classes. Apprentice scribes familiar with it from repeated performances may have written down favorite passages, and occasionally improvised lines for their own amusement or to relieve their study of Sumerian poetry (18). Notice that there is nothing in this scholar’s reconstruction of this stage to support Janko’s contention of a wide circulation of written NE epic. The role George assigns to performance seems justified by the presence in the OB version of much material that had no place in the Sumerian canon (20). Although he does not defend the alleged role of performance on the ground of stylistic features supposedly diagnostic of oral telling, which an archaizing mannerism might imitate, he avers that “it is to my mind inconceivable that ancient Mesopotamia was without traditions of oral poetry throughout its long history, both because the majority of people in all periods could not read or write and in the light of the strong traditions of oral literature in the more recent Near East” (21). [48] George refuses to speculate how the oral material came to be written. He observes the obvious: a literate people who wanted to leave a record for posterity would naturally resort to writing (22). I might add that the deeply rooted cultural practice of archiving and the possession by individual scribes of personal libraries might serve to motivate the act of recording. [49]
In the late Bronze Age (during the MB period), when Akkadian was the lingua franca, cuneiform writing was in much demand in the chanceries of Syria, Palestine, and Anatolia, and, for a brief time, even in Egypt’s El-Amarna. Babylonian texts, including the OB Gilgamesh, were widely copied by scribal schools throughout the West. Tablets have been found at Emar (southeast of Aleppo, on the Euphrates), Ugarit (on the coast of Syria), Megiddo (in Palestine), and Boǧazköy (in Anatolia) (24). [50] The borrowing of Mesopotamian literature was gradual, over several centuries (26). Babylonian high culture had an impact wherever cuneiform was used. South Mesopotamian culture had already been exported to the West (Ebla, Mari, and Tell Beydar) in the early Bronze Age, and archaeologists have found an Old Assyrian pseudo-autobiography of Sargon in a merchant’s library in Cappadocia dating to the early second millennium. Babylonian literature influenced Anatolian scribes in the Old Hittite period (seventeenth and sixteenth centuries), and thereafter continued to make its way into this area directly or through the Hurrians (27). Once again, diffusion is driven by scribal training and the spread of cuneiform literacy. Some intellectual exchange must have accompanied regular trade (as witnessed by the merchant mentioned above); but the educated man, not a scribe, who could read and wanted to own a literary text in cuneiform, will have been the exception. The last centuries of the MB period saw the organization of Babylonian literature into canonical series. MB tablets from Nippur show the Akkadian Gilgamesh as a text the scribe was to face early in his training (35).
We now reach the Neo-Assyrian period, the most immediately relevant because it overlaps with the derivation of the alphabet in Greece and Janko’s dates for the dictation of the Homeric poems. Archaeologists have discovered a library that belonged to a family of ‘chief singers’ (nargallu) from seventh-century Aššur. This seems to support the oral performance of traditional narrative poems. The collection of tablets is “a fairly typical example of a first-millennium private library” (34); it contains school tablets, various documents, and common texts from the scribal tradition with an atypical emphasis on hymns and mythological poems. Which texts these self-styled singers performed, and whether they sang or recited, is not known. Unclear, too, is the connection, if any, these texts have with court performance, and whether Babylonian narrative poetry was still living in the mid-first millennium (35). Not much is known about scribal training in Babylon during that time, although a study suggests an initial phase of exposure to a small selection of Gilgamesh and other celebrated poems. It is possible that this pedagogic practice attempted to build upon the early familiarity of Babylonian children with oral versions of this reduced canon (36). Syrian and Anatolian scribal centers, on the other hand, reveal the use of Gilgamesh at an advanced stage of the curriculum (35). The Babylonian and Assyrian social circles that possessed written literature during the first millennium (usually in personal library collections built over several generations) included scribes, diviners, exorcists, and cult singers, for whom literacy was necessary. Common are colophons which identify the inscriber of the tablet as the young relative of the owner and explicitly label him as an apprentice or junior professional (37). Literary transmission across generations was restricted to a few professions and it took place within families devoted to the given trade, typically, from father to son. This does not mean that Gilgamesh was necessarily unpopular among the educated outside scribal circles. The use of some of its episodes in Mesopotamian art speaks for a wider currency of the legends of Gilgamesh, if not of the poem that bears his name. It seems to me, however, that not much can be advanced with any certainty about its diffusion beyond scribal circles. Writing, collecting, and archiving were immediately, and most intimately, connected with the scribal trade. Apart from these habits of scribal families and the archival and collecting practices of rulers, there is little evidence that, upon hearing the oral performance of Gilgamesh, the average literate man would feel the need to acquire a copy for his private use. But one must assume some such scenario as a regular happening, if incidental exposure to NE literary culture (perhaps mediated by Anatolia or the Levant) was to suggest to a wealthy Greek ruler the value of owning a written record of an extensive Homeric performance. The logic and method of such a trans-cultural influence remains as inexplicable to me as the assumed scenario is implausible. George affirms that West’s theory of an early seventh-century “hot line” between Assyrian court literature and the Iliad (West 1997:627) “supposes that the written epic was put to use as an entertainment in the Neo-Assyrian court. This is an assumption that cannot be proved” (56). Mesopotamian popular literature runs in parallel with scribal high culture but we know nothing about it. Some version of Gilgamesh must have been sung even after the written texts reached their several fixed forms. George thinks it unwise to assume that oral performance must have faithfully followed a written version: “Those that could read may have used the written text to refresh their memories, but others that could not probably knew by heart a version of the poem at some remove” (56).
Someone may adduce the alleged presence of NE motifs and narratives (or narrative patterns) in Homeric and Hesiodic poetry as evidence that Greek singers were acquainted with the written texts of corresponding NE poems. If so, the choice to write down a Homeric performance might not be, after all, so great a departure from the oral culture of Greece. But George makes two observations that helpfully counter this argument. First, that in the study of trans-cultural influence, one should start with Phoenician and Aramaic literature, which by proximity and contact have a prior claim over NE Babylonian or Assyrian literature (56). This draws attention, once again, to our need to investigate the extent and nature of Phoenician literature. [51] Second, that NE oral as well as written composition often consisted in adapting and stitching together well-known motifs and mythologems, sometimes reusing blocks of existing lines. The intellectual exchange that often accompanied commerce would have made such familiar episodes, standard passages, and staple motifs and narrative patterns the common store of singers and poets composing in the various ancient NE languages (56). Greece, too, would have imported from the eastern Mediterranean such “motifs, episodes, imagery and modes of expression that were always traditional in the narrative poetry of the area or had been adopted into that poetry from Mesopotamia long before” (57). In sum, the theory that the dictation of the Homeric poems followed the trans-cultural model of written circulation characteristic of NE epics fails to persuade: first, because the written circulation alleged is either doubtful or else of a very different character than what is required to support the claim; and, second, because, even if the type of circulation imagined by proponents of this theory had actually existed, there was no tenable cultural vector to bring the requisite influence to bear upon Dark-Age and archaic Greece.

1.5 Written Phoenician Literature

In 1990 Janko had stated that “the impetus to taking down a text like [the Iliad or the Odyssey] can only have come from the Near East, where written epics were circulating widely” (Janko 1990:330). By 1998, however, the terms of comparison had changed: “One influence on the person responsible for recording must have been knowledge of the existence of written literature, which means the written epics of the Levant” (Janko 1998a:12). At this point, Janko cites Burkert (1992:114–120), who does not, however, review any epic from a Levantine culture that might have circulated in writing during the time when we might reasonably expect it to have exerted the alleged influence. Instead, Burkert considers Mesopotamian, Hittite, and Ugaritic literatures. The latter two belonged to civilizations that were defunct by the turn of the first millennium, whereas with the first we are back to the Near East. [52] The reason is clear: there is hardly any Phoenician literature extant, let alone epic. Scholars (especially classical scholars) assume that there was an abundant Aramaic and Phoenician literature, but that it has been lost because it was written on leather and other perishable materials. Even granting that, as a religious text of first importance, the preservation of the Jewish scriptures is a priori more likely than the survival of other writings not so intimately tied to religious cult, the complete disappearance of early (i.e. pre-Hellenistic) Phoenician literature is hard to understand if it was so vital and copious as assumed. For Aramaic literature, the only surviving work is the Elephantine Ahiqar, cited by Burkert (1992:32) as proof for the transmission of Mesopotamian literature via Syria to Palestine and Egypt. Burkert (like West 1978a:13 before him) assumes the ultimate Mesopotamian origin of Ahiqar, but this is far from established, and modern scholars consider more likely that both the narrative and the sayings were composed in Aramaic. [53] The Book of Ahiqar, then, does not imply the probable existence of a flourishing book market of Mesopotamian (or Levantine) epics that might have prompted the dictation of the Iliad. [54] Scholars since Wendel (1949:93–94) have conjectured a body of Aramaic literature on bookrolls as the link through which features of the cuneiform book format (like the colophon) made their way into Greece. [55] But Wendel supported his theory on the disputed belief that Ahiqar’s Assyrian setting betrayed its Assyrian origin. Moreover, the Sultantepe library of clay tablets suggests that Aramaic-speaking communities with interest in Mesopotamian literature enjoyed this literature in its original languages. [56] We must also bear in mind the warning in Rollston 2008:67–71 not to assume that the linear-alphabetic northwest Semitic writing system was attended by a higher rate of literacy for the Levant of the Iron Age than the far more complex cuneiform script. [57]
But before reviewing our present knowledge of the nature and scope of early Phoenician literature, it is worth pausing for a moment to probe the validity of the ‘influence model’ espoused by these scholars. Why should we expect a Greek who had come across a Levantine or Near-Eastern epic in written form to think of producing a transcript of the performance of a traditional oral rhapsode? This is an odd notion, rarely examined, which does not withstand scrutiny. The expectation of a cultural borrowing is only reasonable where the borrowed practice meets an existing or incipient need. I do not believe that the scenario in question meets this standard. To put it in terms of a colorful illustration: an inhabitant from a temperate climate who visits the Eskimo would hardly think of borrowing the construction of igloos to meet the need for housing in his own culture. The material, ice, would be scarce or unavailable; the resulting structure would not be durable; and, presumably, his society already has in place adequate building practices to meet the need. So also in the present case. The song culture of archaic Greece was oral. Epic poetry was composed in performance with traditional language and traditional themes; and, crucially, it was preserved and transmitted orally: neither audience nor performer can be plausibly expected to have felt the need to preserve a particular performance, or a fortiori a poetic tradition, in writing. The festival calendar, with its seasonally repeated performance venues, gave cultural expression to the enduring character of its songs. Why should acquaintance with a wholly different method of preservation, well suited for a culture built on the foundations of scribal schools and scholars, [58] public and private libraries, extensive administrative archives, and habits of collection [59] —a method, that is, well motivated by, and readily intelligible within the context of, such a culture—prompt a member of the archaic Greek song culture to adopt the alien and culturally inorganic practice? In archaic Greece, the skills and materials needed for the feat of writing the Iliad must be assumed as scarce or lacking as ice would be for our hypothetical visitor to the Arctic. And the resulting artifact—a small library of leather or papyrus scrolls, according to some—would be culturally marginal and as unlikely to endure as a structure of ice in the tropical sun. It would neither be used by performers, who would continue to observe their age-long performance practice, nor read by a public that, even if it was not largely illiterate, would feel no need to experience in writing what it had hitherto experienced in performance.
Unsuccessful too are the attempts to motivate the borrowing of the alien practice of writing by speculating that it met the emerging need to legitimize monarchic or aristocratic ideology at a time of social turmoil. They flounder in trying to establish that the ideological cast of the Iliad or the Odyssey is, as the case might be, monarchic (Janko 1998a:13) or aristocratic (Morris 1986:122–126). The attempt to pin down the ideological texture of the poetic tradition relies on the same assumption it is called to support: that the poems were written down substantially as we know them at a point in time in order to advance a particular social agenda. Rather, our analysis must accommodate the fact of recomposition in performance with varying degrees of fluidity during a protracted interval that included the archaic period. This acknowledgment allows the diachronic dynamics of reception to come into focus. The pivotal development and determinative context for this reception is the rise of the polis, with all of its evolving ideological complexities. Proponents of these theories rarely make the effort to explain in detail why the written artifact should actually advance the ends of a monarchy or an aristocracy. Morris (1986:122–126) provides an exception to the rule; I explain below [60] why I find his reasoning unconvincing. Although the argument is not always or necessarily made, it is worth noting that writing per se does not favor monarchic, aristocratic, or democratic tendencies, although arguably, depending on the particular social context, it can be used to further any of these political arrangements.
With these preliminary observations in place, I now proceed to survey with Krings 1995 as a guide what is known about Phoenician literature, and to assess whether Burkert, West, and Janko are justified in making it at least partially responsible for the writing of Homeric poetry. It is important to heed her opening warning: the significance of the Phoenician alphabet for Western culture does not eo ipso justify the expectation of a commensurately significant Phoenician literature (31). Scholars like Lipiński 1992 have approached the subject largely with generic categories derived from the study of Greco-Roman antiquity. More problematic yet are: the naive inclination to take at face value ancient testimonies about the nature and extent of the Phoenician and Carthaginian literatures; the assumption that what applies to latter-day Carthage must also be applicable to earlier Phoenicia; the failure to consider the possible influence of Greek literature on Carthaginian and later Phoenician works; and the interpretive weight assigned to Pliny’s problematic testimony about the existence of a library at Carthage (Naturalis historia 18.22–23). [61] Krings (1995:32) pronounces flawed and unfruitful the approach that seeks to argue for the existence of a Carthaginian literature from Greco-Roman sources that make reference to Carthaginian works.
To take the nature and scope of Phoenician literature first, it is the case that many scholars assume that it was rich and considerable. According to Josephus in Against Apiōn 1.28, all acknowledged that significant administrative and historical record-keeping distinguished the Phoenicians. We should mark the obvious fact that the writing in view here concerns historical records, not ancient epics that might have prompted the dictation of Homer. But, as it happens, what Josephus declares an uncontested fact is far from certain. The sources he alleges [62] are all suspect in various ways. [63] Although the existence of the alleged Tyrian chronicles cannot be ruled out, it is impossible to ascertain their nature with any confidence. [64]
Philo of Byblos claimed that his Phoenician History, which survives in fragments quoted or paraphrased by Porphyry and Eusebius, derived from a Phoenician scholar from before the Trojan War by the name of Sankhouniathon. [65] This erudite Phoenician from the latter half of the second millennium BC is said to have collected and interpreted the cosmogonic writings of Taautos (the Greek Hermes!) stored in the adyta of the temples of Ammon. [66] Baumgarten observes that the fragments of Philo’s History “are virtually the sole testimony to the myths and beliefs of the Phoenicians” (1981:262). Thus, for the last two hundred years, the existence of an early Phoenician literature has hinged primarily upon the historicity of Philo’s source and his early date (or, in lieu of this, the early date of his source material). [67] Baumgarten (1981:264–265), Lemaire (1986:217), and Krings (1995:33) all agree in their judgment: in all probability, behind the myth of Sankhouniathon stands a collection of sources of Phoenician cosmogonic lore from the Hellenistic period, redacted in Phoenician or, more likely, Greek. These had been gathered together under the venerable pseudonym, or else circulated individually and were later compiled by Philo himself. [68] This raises the question of the antiquity and authenticity of his sources, and—of obvious importance to my argument—of their character, oral or written. For Baumgarten, Philo’s mention of Mōt (806:21), the Zophasēmin (806:24), and the poetic form of the cosmogony and zoogony establish beyond doubt that their ultimate source was Phoenician. [69] A review of the last argument is instructive. Philo’s sources for the cosmogony and zoogony appear to use the poetic device of parallelism familiar from Biblical and Ugaritic literatures. This conclusion, however, is not without its problems, since there are sentences that lack the expected parallelism and others whose alleged parallelism presents alternative explanations. [70] Baumgarten makes two important comments at this point. First, the use of this poetic device does not rule out Greek influence since the Jews employed parallelism in poetry clearly influenced by Greek thought, and even in poems written in Greek. [71] Second, nothing in the syntax or style prevents a date later than the eighth century BC. [72] He also believes that Greek science explains aspects of the cosmogony previously inexplicable, and that its source must therefore date to a time after the sixth century BC and had Greek sources: [73]
The Hermetic and Gnostic texts cited establish … that Philo’s cosmogony as interpreted here was not sui generis. Not only do we have Eudemus’ report of a similar Phoenician cosmogony from the fourth century B.C. at the latest, but the neighbors of the Phoenicians were also engaged in reinterpreting their traditional doctrines in the light of Greek philosophy. … [I]t is tempting to suppose that Philo’s source was written in its final form at approximately the same time as C[orpus] H[ermeticum] Ⅲ and the cosmogony of the Nicolaitans. A date at the very end of the first century B.C. or during the first two centuries A.D. seems proper. This would be close to the time when Philo himself was active.
Baumgarten’s judgment, then, is that Philo’s work contains genuine Phoenician traditions but does not preserve one or more Phoenician texts from hoary antiquity. [74] One cannot infer from Philo with any confidence the wide circulation during the archaic period of a rich and varied collection of Phoenician myth, heroic or cosmogonic, in written form. I do not doubt that, like the civilizations around them, the Phoenicians too had their own myths and corresponding social contexts, private and public, for their telling. But in the absence of compelling evidence the presumption that these myths were composed in, or reduced to, writing and circulated primarily or exclusively as scrolls seems to me unwarranted. I find no convincing rationale to look here for the stimulus to record the Homeric poems in writing.
There is one last piece of evidence from which scholars have sought proof of the existence of epic narratives among the Phoenicians: a silver bowl of Phoenician craftsmanship from ca. 710–675 BC found in a tomb at Praeneste. [75] Another bowl in a poorer condition from Kourion, Cyprus, displays a similar sequence of scenes. [76] West (1997:99) speculates that the story on these bowls “may have been the subject of a Phoenician poem.” [77] This is not a safe conclusion. The time is long past when pictorial narratives on Greek vases were commonly assumed to be derivative of particular poetic compositions. [78] Vase artists should not be assumed merely to respond to the initiative of singers. In archaic Greece, the domains of song and of pictorial representation belonged to the same traditional cultural matrix. Both drew on the cultural superset that the French call ‘l’imaginaire mythique,’ which exists as a symbolic lexicon and a narrative grammar, together with a collection of ever-changing multiform narrative instantiations. [79] Within their respective spheres of work, singers and pictorial artisans both responded to the forces of reception with calibrated mixtures of tradition and innovation. [80] Just as their works faced similar cultural forces, either domain evolved diachronically under the influence of the other. Although we know little about the corresponding Phoenician domains of poetic song and pictorial representation, it is naive to assume the unidirectional influence long abandoned as fallacious in the case of archaic Greece. The artisan of the silver bowl may have known how to depict a traditional religious motif (let me call it ‘divine deliverance of the pious ruler from the attack of a savage’) without the aid or inspiration of a poetic narrative archetype. But let me agree for the sake of argument that the silver bowl depicts a poem: why should it be an epic and not a brief composition of some other, culturally appropriate, Phoenician genre? And, the more important point for my argument: why should we further assume that this composition circulated primarily, or exclusively, in written form? [81]

To return to Janko, I will close this section by observing that his exposition (in Janko 1998a:13) of the ideological motivation for the dictation of the Homeric poems seems unlikely to survive scrutiny. How precisely are we to imagine that the written artifact, as opposed to the performance of the poem, buttressed the legitimacy of monarchic rule? Retaining the services of ‘Homer’ for frequent performance of the poem would seem more effectively to serve such propagandistic ends. The leadership of Agamemnon, moreover, often portrayed as flawed and perplexing, and Akhilleus’ brazen challenge to his authority, central to the poem, seem hardly counterbalanced by sentiments such as Α 277–279 and Ι 95–99. If legitimizing ‘monarchic’ rule (however defined) was the end, many an episode might seem best left out.


[ back ] 1. Janko 1998a provides a helpful review of the evidence and defends the oral nature of the poems’ origin. As I explain below, however, I cannot agree with Janko’s dictation theory. Jensen 2011 provides a full-scale comparative review of the ‘Homeric Problem’ in light of modern fieldwork on oral epic traditions.
[ back ] 2. I will usually refrain from placing Homer in scare-quotes. The reader should bear in mind that scholars’ use of this traditional name does not necessarily imply their acceptance of ancient biographical traditions or even that this was the name of the singer who is often called the “monumental composer.”
[ back ] 3. Reece 2005 reviews and defends oral-dictation as the most plausible mode for the recording of the Homeric poems. Although useful in various ways, his summary is not reliable when it mischaracterizes the evolutionary model, implying that it advocates unchanged and unabated fluidity in the degree of recomposition in performance during the classical and Hellenistic periods (so at 77 and 84). Reece also caricatures the model when he intimates that, if true, one should expect the Homeric epics to be “simply a collection of loosely related episodes, … the predictable result of a process of compilation by various hands over a long period, or of a process of gradual accretion within an impersonal oral tradition” (56; cf. 65). As to Atticisms in the Homeric poems, which he surveys in 80–86, much more can be said than he does there, including evidence that he overlooks. But a comprehensive—and necessarily technical—answer to Reece’s challenge (84) must wait for another occasion. I maintain, however, that on balance the data, objectively viewed, support the true claims of the evolutionary model and hardly constitute an Attic dialectal “veneer.”
[ back ] 4. I am using ‘paper’ generically for any potential substrate.
[ back ] 5. Lord 1991:43.
[ back ] 6. Janko 1998a:3, with reference to so-called “transitional texts.” But see the adjustments suggested by Jensen (1980:90), who prefers “oral autograph” for such hypothetical written texts.
[ back ] 7. In keeping with its original linguistic meaning, I label ‘diachronic’ those aspects that follow from the development of the poetic tradition as a literary and linguistic system. ‘Diachronic’ encompasses the outworking of systemic tendencies. ‘Historical,’ on the other hand, denotes all that took place or is assumed to have taken place. When opposed to ‘diachronic,’ ‘historical’ refers to what does not flow from, and cannot be predicted on the basis of, the evolving tendencies of the system. Diachrony must be open to the corrective of history, for accidents cannot be excluded a priori. But, in the absence of strongly supporting evidence, the scholar who engages in historical reconstruction cannot expect conjectured historical accidents to overrule with plausibility otherwise anticipated diachronic developments.
[ back ] 8. First proposed in 1981, it is found most conveniently in Nagy 1996b:26–63, and at greater length in Nagy 1995.
[ back ] 9. Nagy 1995:169 and Nagy 1996b:52.
[ back ] 10. If wide attendance at first helps to ground the reputation of the festival, once its prestige has solidified it no longer needs a pervasively supralocal audience, or even a majority of supralocal performers, to continue to exert its influence. At that stage, the tastes of a largely local festival audience may be considered normative and cast a long shadow over the performance of the given poetry elsewhere.
[ back ] 11. See Allen 1924:283–296; van der Valk 1949:14–21; Citti 1966; Janko 1992:26n29; Nagy 1996c:147–148; and Cassio 2002:117 and n. 58. For the aesthetics of common reception, see “koine (Homer)” in the index of Nagy 1996c.
[ back ] 12. This is Nagy’s “definitive period” (1996b:42).
[ back ] 13. Regarding the claim that Parry articulated a dictation theory of his own, see Nagy 2003:66–67.
[ back ] 14. Lord 1991:44.
[ back ] 15. Lord 1991:44. To the suggestion of an external impetus by an appreciating audience “who recognized the special merit of the Iliad,” one may oppose Sealey’s comment that “as far as can be discovered, those Greeks had learnt to recognize merit, not in songs, but in singer” (both quotes are from Sealey 1957:330).
[ back ] 16. So Nagy 1996b:32: “[H]ow exactly was such a dictated text supposed to be used for the process of performance? How could a dictated text automatically become a script, a prompt, for the performer who dictated it, let alone for any other performer?”
[ back ] 17. Cf. Lord 1991:45. This is why it is not, in fact, paradoxical at all, pace Cassio 2002:107, that one “who had explored the techniques of oral composition in more depth than anybody else” supported a culturally marginal dictation model. Furthermore, it is hardly true of Lord’s proposal “that the transcript became an indispensable basis of all subsequent recitations” (107).
[ back ] 18. Hence the comments by Sealey 1957:329–330.
[ back ] 19. Lord 1960: “One should emphasize, however, that these changes or differences are not caused by the singer’s conscious or deliberate choice of an order of words or of words themselves for any other reason than the influence of the surrounding rhythmic structure. … The singer is struggling with the traditional patterns under unusual circumstances … ; he is, indeed, striving to maintain, not to depart, from the tradition” (127–128). Cf. Lord 1991:11.
[ back ] 20. Wilson 1970:169. Cf. Koljević 1980:314–318.
[ back ] 21. The implausibility would hardly be relieved by assuming an autographic text instead.
[ back ] 22. Cited in Koljević 1980:322.
[ back ] 23. In [Plato’s] Hipparkhos 228b and Diogenes Laertios 1.57. More on this below, §
[ back ] 24. Cf. Ford 1997.
[ back ] 25. For an Indian epic comparandum, cited by Nagy 1996b:77, cf. Blackburn et al. 1989: “[W]hen an epic story is well-known to the audience, the complete story, from beginning to end, is rarely presented in performance—or even in a series of performances. The full story is sometimes found in written and published texts, but we prefer to speak of an epic tradition that encompasses not only text and performance but also what is unwritten and unperformed” (11).
[ back ] 26. Janko 1998b:207; cf. Janko 1990:329–330.
[ back ] 27. “Even the most naively optimistic scholar may feel uneasy when invited to believe that what we have happens to be just what Homer sang … . But where the former theory [that there were no written texts until the middle of the sixth century] founders … is in its inability adequately to account for the early textual fixation implied by certain details of the linguistic constitution of the poems. Janko’s comparative statistical study … shows that the Homeric poems’ linguistic evolution was arrested at a very early point.”
[ back ] 28. On which see below, §2.1.
[ back ] 29. Related concerns are expressed by Clay 1997:490–492 and Burgess 2001:52–53. For Janko’s disapproval of West, see Janko 1998a:1.
[ back ] 30. Jones 2010:290; cf. Jones 2008.
[ back ] 31. I single out this particular instance because Janko’s recurrence to it hints of its special significance to him.
[ back ] 32. Χ 335 is a special case: either proleptically, of Hektor’s impending death, or stricto sensu of the collapse of his limbs (ἤριπε δ’ ἐν κονίῃς 330).
[ back ] 33. Note in particular Odysseus’ reasoning at σ 90–94. A comparable ambiguity, this time between death or love as the loosening agent, seems in view at ξ 69.
[ back ] 34. Leaf 1900–1902:2.33 ad Ν 420–423 remarks that “it would be quite unlike the Epic style to represent him [i.e. Deiphobos] as mistaken without explicitly saying so.” I do not find the force of this sentiment decisive.
[ back ] 35. Nagy 2004:114 observes that στενάχοντες occupies the same verse-final slot at ξ 354, to which I add that μεγάλα, which precedes στενάχοντες there, is a variant of βαρέα in βαρέα στενάχοντα at ψ 317. For the notion of the (Byzantine) ‘vulgate’ text, see Janko 1992:20–22.
[ back ] 36. And perhaps not have two others leave the battlefield to take him back to the boats.
[ back ] 37. Cf. Nagy 2004:113–115 for a somewhat different solution. As Nagy observes, how plausible is Janko’s view that for half a millennium Ν 402–423 “happily coexisted with the version of Ν 423 that featured στενάχοντα—until Aristarkhos in the second century BCE finally offered his ‘solution’”?
[ back ] 38. That Janko does not grapple with this psychological implausibility seems incongruous with his perplexity when he considers Nagy’s evolutionary model: “[O]ne is entitled to ask why the resulting texts contain so many minor oddities, which would surely have been tidied up in any process of this kind” (Janko 1998a:12n63).
[ back ] 39. The historian notes that early Ionians used leather; Janko adds, “as did the Phoenicians and Arameans” (1998a:12n63).
[ back ] 40. This, again, is the distinction between the diachronic—what one might expect from the natural development of culture as a system—and the historical, which might have room, however implausible, for an anomalous, early application of the alphabet to a novel use for which it had not been created.
[ back ] 41. Cf. Csapo et al. 2000:103–105. Johnston 1983:67 adds a concern for the “personal marking of property.” That proprietorial and commercial interests encouraged the use of writing is not only the most likely, but also the consensus, view. So, for example, Ruijgh 1995:37–38: “[I]l est légitime d’admettre que l’usage de l’alphabet grec, adopté en vue de la comptabilité commerciale, a été limité à l’administration économique pendant plusiers siècles” (38). Cf. Woodard 1997a:252 (and n. 15) regarding the minority view of Powell 1991 that the alphabet was created to record Homeric poetry: “[A] writing system … expressly engineered for such a high-minded and noble purpose as recording poetry … seems not altogether probable.” This theory is also contradicted by the failure of the early scripts to distinguish long and short vowels, a feature that is of central prosodic significance (cf. Woodard 1997a:243n111 and 253). To the notion that, phonologically, a syllabary was too indeterminate “to have served as a practical vehicle for recording ambitious poetic compositions” (Powell 1991:113), Woodard 1997a:254 replies that “the syllabic script of Cyprus was an effective means for recording the Greek language and, as such, could certainly be used for writing verse compositions. Powell has confused language with script” (cf. Consani 2008:154–155, who underlines the natural agreement of syllabic systems with “speakers’ spontaneous analysis of [the] linguistic continuum, as it is operated when writing a text”). Robb 1994, like Powell, doubts that the alphabet was designed for commercial purposes. Its initial motive was “to record the hexameters of dedications” (8). His argument is circumstantial and turns on the character of the earliest surviving inscriptions and it is vulnerable to the same objections levied against Powell’s. It is true that archeology has not unearthed any alphabetic commercial records. But it is surely for these, not for the monumental Homeric poems, that we expect the use of perishable and easily reusable substrates that would not have survived. If the Linear B tablets had not been preserved by the fires, who would have suspected that the Mycenaean palaces had kept administrative records? A recent opponent of the commercial origin is Schnapp-Gourbeillon 2002:255–314. She allows Iron-Age Euboian aristocrats both the practice of commerce and the use of writing (279). But even though she owns that writing must have remained marginal vis-à-vis performance (296), her treatment is culturally simplistic at key points. One example must suffice: “[O]n est conduit à penser que l’impulsion menant à la rédaction de cette oeuvre ne s’est produite qu’en raison de sa qualité, son originalité extrême, tellement au-delà du reste de la production épique que l’obligation d’écrire s’est pratiquement imposée d’elle-même” (294).
[ back ] 42. Even the assumption that papyri might have been used faces significant obstacles, technical and economic. (Janko 1998a:12 recognizes the need for considerable time and resources.) For the availability and use of leather and papyri, see below, §3.
[ back ] 43. Following the lead of Burkert (1992:25–33), West (1988:170) offers a different reason when he traces the impetus back to the Greeks’ contact with the Phoenicians and the Aramaeans from North Syria. (Among the Greeks, the Euboians are principally in view, with a possible role for Cyprus.) Perhaps under the influence of the intervening Burkert 1992 and West 1997, Janko 1998a:12 abandons his 1990 reference to the Near East in favor of “the written epics of the Levant,” singling out “the written literature of the Phoenicians” in 12n64 (cf. Burkert 1992:32–33). The existence and character of a written Phoenician literature is an important matter I take up below, see §1.5.
[ back ] 44. As George 2003:39 notes, of the great narrative poems only the Enuma elish exceeds Gilgamesh in the number of first-millennium sources.
[ back ] 45. Excluding Tablet Ⅻ, which most scholars agree does not belong in the series: “The eleven tablets of the epic vary in length from 183 to 326 lines of poetry, so that the whole composition would originally have been about 3,000 lines long” (George 1999:xxⅷ).
[ back ] 46. Cf. Tigay 2002:xxⅱ.
[ back ] 47. Shulgi’s dates are 2094–2047 BC in conventional chronology.
[ back ] 48. Cf. Vogelzang and Vanstiphout 1992.
[ back ] 49. This seems to be the source of the so-called Pennsylvania (CBS 7771) and Yale (YBC 2178) tablets, fine specimens whose quality suggests permanent library copies (cf. George 1999:xx).
[ back ] 50. They were also composed in Hittite and Hurrian (George 2003:24n66).
[ back ] 51. On which see immediately below, section 1.5
[ back ] 52. Cf. Burkert 1992: “The establishment of the first Greek library—the Iliad written down in twenty-four(?) leather scrolls—and of the great library of Ashurbanipal at Nineveh, who ruled from 668 to 627, may well have taken place at about the same time. Even this may not be totally accidental” (120).
[ back ] 53. Cf. Vanderkam 1992:120.
[ back ] 54. Neither does its sapiential cast commend the notion that a written copy of this very work might have provided the impetus for the recording of Greek heroic epic.
[ back ] 55. Cited with approval by Burkert 1992:32. Cf. West 1997:592–593.
[ back ] 56. Cf. Gurney 1952.
[ back ] 57. Cf. Stoddart and Whitley 1988, who highlight the complexities that attend the study of alphabetic literacy. Their study compares the disparate uses and ideological meaning of alphabetic writing in Attica and Crete, and in Greece and Etruria.
[ back ] 58. Cf. Driver 1976:62–73.
[ back ] 59. Driver 1976:73–77.
[ back ] 60. See p. 44 n. 16.
[ back ] 61. All of these factors are carefully reviewed with commendable skepticism by Krings 1995. She addresses further the influence of Greek literature in Carthage in Krings 1991.
[ back ] 62. ‘Tyrian chronicles’, γράμματα δημοσίᾳ γεγραμμένα (Against Apiōn 1.107), mediated by an otherwise unknown Dios (FGH 785); a Phoenician history by Philostratos (FGH 789); and Menander of Ephesos.
[ back ] 63. These sources are listed, with brief commentary, in Krings 1995:33.
[ back ] 64. Cf. Lemaire 1986:217–219.
[ back ] 65. The dating by the Trojan War is Porphyry’s. By making him Hesiod’s source (Baumgarten 1981:89n89), Philo merely makes Hesiod a terminus ante quem.
[ back ] 66. Baumgarten 1981:10–11, with translation in 63–64.
[ back ] 67. Krings 1995: “La date et l’historicité de ce Sanchouniathon ont été souvent discutées et sa prétendue oeuvre s’est retrouvée très tôt au coeur d’une bataille d’érudits, révélatrice de la sensibilité persistante des débats sur la ‘littérature’ phénicienne” (33). For bibliography on the debate, see Lemaire 1986:233n19.
[ back ] 68. Baumgarten (1981:264) favors the notion that there were many sources and that Philo has combined the traditions of at least two cities, Tyre and Byblos. Each source faces the student with dating uncertainties. West (1994a:294n20) remarks that “Sanchuniathon is a genuine Phoenician name (Šakkūnyātōn, ‘Shakkun has given’), but of a type not likely to be earlier than 700 B.C.”
[ back ] 69. Baumgarten 1981:96.
[ back ] 70. Baumgarten 1981:98–99. Note also the very different explanation in West (1994a:296) for what Baumgarten calls parallelism.
[ back ] 71. Baumgarten 1981:102 and nn. 27–28.
[ back ] 72. Baumgarten 1981:103.
[ back ] 73. Baumgarten 1981:122. The passage cited comes from page 128. Cf. West 1994a.
[ back ] 74. Baumgarten 1981:265.
[ back ] 75. Rome, Museo Archeologico di Villa Giulia, Inv. 61565. Cataloged by Markoe 1985:191 as E2, with photographs in 278–283. A good color photograph can be found in Gehrig and Niemeyer 1990:37, with a description in 186–187.
[ back ] 76. Metropolitan Museum, Inv. no. 74.51.4556. Cataloged by Markoe 1985:177 as Cy7, with photographs in 254–255.
[ back ] 77. He is not the first to do so. Called the “Ape Hunt episode” by Markoe 1985:49, he further notes that “the specificity of the theme, the cohesiveness of the story line, and the element of divine intervention … argue strongly for the supposition that the story itself is not merely the product of a Phoenician artist’s vivid imagination but does, in fact, describe a lost fable or epic” (67–68). The reader should ponder these words in the light of my comments immediately below. Note especially Markoe’s prejudicial conviction that without the aid of a verbal narrative the artist could not have produced a compelling, cohesive pictorial narrative. Following Clermont-Ganneau 1880:16, Güterbock 1957:69 had characterized it as a “stage play.” For the Kourion bowl, see Marquand 1887.
[ back ] 78. Cf. Goldhill and Osborne 1994:1–11; Hedreen 2001:3–12; and Small 2003.
[ back ] 79. For Greek pictorial grammar, see Stansbury-O’Donnell 1999; Hedreen 2001:12–18; and Woodford 2003:15–102.
[ back ] 80. Cf. Woodford 2003:105–140.
[ back ] 81. Even if the bowl had been inspired by poetry, it need not have been Phoenician poetry. Produced in Cyprus at the intersection of Greek and oriental cultures, the artist might have used the oriental pictorial motifs to depict a Greek poetic narrative. This is precisely the opinion advanced by Burkert 1992:104. West 1997:100–101 counters that the female sun that rescues the hunter makes clear that the story is of West Semitic origin. The argument is not persuasive: a Phoenician artist might readily translate a Greek narrative into his own artistic conventions. Furthermore, the interpretation of the winged disk is rather fraught, as Clermont-Ganneau 1880:89–92 and Matthäus 2005 show.