In response to:

Becoming Homer from the March 5, 1992 issue

To the Editors:

Reading Hugh Lloyd-Jones’ review of Albert Lord’s Epic Singers and Oral Tradition [NYR, March 5], your readers might get the impression that the idea of oral composition of the Homeric epics is a theory that will die as soon as a few hold-outs come to terms with the facts. I hope I can show that more supports this notion than simple academic inertia.

One may as well start with the general notion of oral poetry and the attempt to illuminate Homer by the study of living traditions. It is correct to say, with Ruth Finnegan (1977), that “oral poetry” is not a single, homogeneous category. At the very least, the characteristics of particular oral traditions vary according to the surrounding technology (especially by interaction with non-oral traditions). Furthermore, the social function of the texts affects their generic shape. For instance, the Vedas (scripture) were apparently handed down orally for many generations with minimal changes, while most oral traditions change rapidly to adjust to societal changes. Oralists such as Paul Kiparsky (1974) and D. Gary Miller (1982a,b) have been aware of this for some time. This variety is not, however, a real objection to the work of Parry and Lord. In fact their claims were too weak: not only does Homer look like extant oral poetry, but it belongs to a particular type of oral poetry also including the south Slavic and Kirghiz (central Asian) epic traditions.

It is worth noting at this point that the word “oral” may distract from the real issue for the literary critic who is in any case working with written texts. The important question for us is whether Homer (whether dictating, writing himself, or even reciting to memorizers) composed under the limitations of the oral poet? This is, did he allow himself to pause in mid-work to refer back to or even change some part that was previously composed? And to what extent did he use words or phrases as metrical filler, rather than for their denotation? Thus I would describe the central issue to be simultaneous composition-in-performance, not orality. I will, however, use the two terms more or less interchangeably since in most cases they go hand in hand.

Professor Lloyd-Jones rightly identifies the question of “economy” as central for deciding the orality of the Homeric epics. This is particularly so since, as Lord (1960) showed, economy is practiced by individual oral poets, but not by whole traditions. Only the pressure of actual experience drives them to minimize their phrasal repertoire. Hence, if Homer can be shown to be highly economical, we will have reason to believe that Homer was an active participant in a tradition of oral composition rather than a later imitator of its style. David Shive’s Naming Achilles (1987) is presented as the death blow to economy and so to the notion of oral composition. But as many of Shive’s reviewers have pointed out, he does not really understand the notion of economy. A rough definition of the principle is “the avoidance of two expressions of the same metrical shape and the same meaning.” Hence, part of the multiplicity of forms for “to Achilles” is the expected variety of metrical forms (a characteristic of much oral poetry called “extension”). Furthermore, much of Shive’s argument depends on a failure to distinguish between meaning and reference. Even if the word anakti “to the king” (Il.24.449) refers to Achilles, it certainly does not mean “Achilles” since the same word is used to refer to, for example, Cryses at Il.1.390. If one imagines a set of five or six such non-synonymous expressions (some as general as words for “he”), each with five or six different metrical forms, one quickly reaches Shive’s purported total of 32 (actually, I count 35) forms for “Achilles” in the dative without any true redundancy. (A similar flaw makes its appearance in the word of Austin, cited p. 54.) In fact, the wide variety of metrically distinct forms that do mean “Achilles” means that Homer does not need recourse to so many of the more general terms. One should also note that Achilles is the central character of the poem and is still the source of very little violation of economy. Paraskevaides’ (1984) examination of phrases for inanimate objects show an even higher degree of economy. Perhaps even more remarkable is Visser’s (1987) demonstration that every one of the 25 expressions for “he killed” in the Iliad has a distinct metrical form.

Visser’s work also provides us with the new insight that Homer composed lines describing certain common actions (such as killing) by what might be thought of as a menu-driven process: The sentence is broken down into a small number of parts, and for each part there is a standardized list of phrases of various metrical shapes. The poet then picks one item from each list. Thus even if none of the individual components are redundant (as they are not), compound expressions of similar meaning can have a variety of forms. Thus many of the forms that Shive (and others) have called uneconomical are not formulae at all, but clumps of formulae.


Economy is not, however, the sole evidence for oral composition. I will mention here only the on-going work of Egbert Bakker, who points out that Homer’s verse tends to fall into small chunks, sized and connected in a manner much more characteristic of oral discourse than written composition. Such cognitive linguistic research may ultimately provide the best evidence for oral composition.

It is objected that oral theory does not allow for the obvious artistry of the Homeric poems. Such objections are often buttressed by quotations from misguided oralists such as Page (cited, p. 53), who seem to deny the evident qualities of the poetry. This is simply not a prediction of oral theory. Consider the fact that ordinary speech is “constrained” by syntax (sentences normally have to have a subject and a verb) and by the limits of the lexicon (speakers of multiple languages often find that there is no equivalent in one for some word in the other). Poets of most eras have been further constrained by the rules of meter. A poet composing by the formulaic method is merely operating with a language which has more (or simply different) constraints than normal. In particular he has a number of large “words” such as swift-footed-Achilles at his disposal.

It is true that anti-oralist criticism and additional field observation have forced a reconsideration of the internal flexibility even of such traditions as that which lies behind the Homeric poems. Lloyd-Jones cites the differences in diction between the speeches and the main narrative of the Iliad. But this difference is much like that between, say, scenes of receiving embassies and of arming for battle. The poet learns a number of traditional type scenes, each with its own set of characteristic formulae. Richard Martin (1989) has even argued both on internal grounds and by comparison with other oral traditions that the speeches can themselves be divided into a number of sub-genres. Lloyd-Jones also notes one of a number of studies which suggest that epithets which traditional oralists had felt to be metrical filler were in fact significant. However, Bakker and Fabbricotti (1991) have shown that while these epithets (and even entire phrases) are usually used in this way in oral poetry, there is no reason that they cannot be used more pointedly in an appropriate context.

More concretely, anti-oralists point to the gap in quality between recorded South Slavic epic and Homer. I am not in a position to defend the former (though this may yet be done), but I would point out that we must in any case posit an extra-ordinary talent for Homer. It is not clear to me that the difference between Homer and a relatively small sample of oral epics is any more to be attributed to differences in technique than the difference between a sonnet of Shakespeare and a Hallmark greeting card.

Lloyd-Jones’ article is, after all, a book review, and it may be the case that Albert Lord’s case for oral composition of the Homeric epics (especially as presented in the work under review) was not airtight. Lord certainly did not respond to the more recent anti-oral criticism Lloyd-Jones cites. On the other hand this does not mean that such a response is not possible. A more upto-date approach to oral poetry is internally consistent and shows that the empirical evidence supports the notion that the Iliad and Odyssey were composed in performance without significant back-reference or editing. Furthermore, the epithets are chosen much more for their general, not local, significance. It is far from settled how the fact of the composition-in-performance of the Iliad and Odyssey should affect their criticism; it certainly cannot be ignored.

Andrew M. Riggsby
Department of Classics
University of California
Berkeley, California

To the Editors:

Was Homer an exclusively oral poet, or did he have recourse to writing? This question has vexed Homeric scholars for a long time, and became especially burning after the publication of the work of Milman Parry and Albert Lord, who not only presented thorough evidence of a living oral epic tradition among Yugoslav singers, but worked out a detailed theory of the techniques of oral composition. Sir Hugh Lloyd-Jones, in his review of Lord’s last collection of articles, has now rekindled this debate. He takes the firm position, as have several before him, that Homer could not have produced such poetry without first attaining literacy, and he buttresses his argument by adducing recent claims (by Barry Powell and Martin Bernal, also reviewed in the same article) that the Greek alphabet could have been invented several centuries earlier than previously supposed, i.e. in time for Homer to have learned to write.


Whether or not the alphabet existed at that time (and without any evidence at all of earlier writing, it strains the mind to think that it could have) is not the central point. Rather, it is Sir Hugh’s fundamental opposition both to the general idea of Homer’s “orality” and to the idea that Parry and Lord could ever have thought of comparing Homeric verse with that of illiterate Yugoslav singers. He seems to view the former as inconceivable (and therefore impossible) and the latter as repugnant (and therefore inadmissible). I find both these views not only limited in scope but also unworthy of the high degree of scholarship represented by Sir Hugh and his school.

The issue at hand is not one of aesthetics but of scholarly responsibility. If one is to express a critical opinion on a piece of verbal art, one should either take the time to learn something of the language in which the art is expressed, or at the least defer to those who do know the work in the original language. Observing correctly that “Lord seems to feel that a person ignorant of Serbo-Croatian has no right to express an opinion” on the artistic value of these poems, Sir Hugh concedes that “a knowledge of that language would certainly make it easier.” The word “it” in his statement seems to refer to his notion of “the kind of thing with which we have to deal,” thus to his intent to prove the Serbo-Croatian material inferior; for he goes on to give his “impressions,” based on summaries and translations, that the Serbo-Croatian verse is both “less effective” and “hardly as distinguished” as the verse of Sir Walter Scott, and that it “seems clear enough that they cannot be compared with Homer.” As do all readers of verse narrative, Sir Hugh has a right to an aesthetic judgment. As a scholar, however, Sir Hugh does not have the right to argue his point by taking material out of context, by failing to take account of relevant scholarship, and by misquoting his sources in a manner that seems almost deliberate.

Sir Hugh takes “a few specimens quite at random,” admitting they are literal (and therefore not poetic) translations. What he does not say is that the first (taken from The Singer of Tales, pp. 109-110) was cited by Lord not to illustrate the aesthetic character of the poetry in general but as a concrete example of father-son transmission. Furthermore, Lord makes no claim that this particular song or singer is directly comparable with Homer. Indeed, in their attempt to discover how oral singers acquired and passed on their knowledge, Parry and Lord studied many singers, often purposefully choosing singers who were “no more than average,” since these were “the type of singer who must carry the brunt of the transmission of the art” (Singer, p. 113).

Sir Hugh’s second example is “from a South Slavic wedding song,” and comes from Lord’s essay on “The Kalevala, South Slavic Epics and Homer” (p. 112 in the collection of articles under review). In citing it, however, Sir Hugh fails to acknowledge Lord’s clear differentiation between “women’s songs” (which include such ritual songs of lament, sung as the young girl leaves home to be married) and “epic songs.” The separation between the two is a basic tenet of South Slavic folk verse scholarship, and no scholar of folk verse will compare wedding songs directly with epic songs. Of course the repetitions in the wedding song “are not at all like those found in Homer,” because they are not part of the epic genre at all! As Sir Hugh notes, songs such as these are more reminiscent of ballad literature, and indeed south Slavists often group wedding songs with ballads. The point is that in no case do they ever consider wedding songs equivalent to epic songs. Glossing over these basic generic differences entirely, Sir Hugh then dismisses “most of the specimens of Slavic poetry quoted by Lord” by calling them ballads. He admits that the two longer Serbo-Croatian poems usually cited in comparisons to Homer (one has over 12,000 lines and other over 13,000 lines) are clearly epic. His “impression” of them, however (again, based on translated summaries and extracts) is that they “cannot be compared with Homer.”

Even the meter of the Serbo-Croatian songs, which he calls trochaic pentameter, displeases him: “I have the impression that it is less effective than the Homeric hexameter.” I fail to see how one can assess the “effectiveness” of a particular metrical form when one has no idea of the effectiveness of the actual words and phrases being communicated in this form. Sir Hugh would have done well to consult the work of John Miles Foley, one of the few classicists who has taken the effort to learn Serbo-Croatian so as to study the Parry-Lord texts in the original. Foley’s recent book, Traditional Oral Epic: Beowulf, the Odyssey, and the Serbo-Croatian Return Song (University of California Press, 1990), among many other achievements, manages to explain the complexity of the Serbo-Croatian verse in its entirety, showing conclusively the utter descriptive inadequacy of the label “trochaic pentameter” for this verse. This work, a masterpiece of comparative epic scholarship, demonstrates convincingly the necessity of studying fully each epic tradition within its own context before attempting comparisons.

The question of orality is more complex. At several points in his essay, Sir Hugh takes Lord to task for not providing better answers to critics of his oral theory, or for not modifying his theory more satisfactorily in light of these criticisms. Indeed, Lord’s oeuvre seems remarkably consistent in its insistence both on Parry’s original principles and on Lord’s own elaboration of these principles based on their collaborative work with the Yugoslav singers. I find this facet of Lord’s work a strength rather than a weakness. The very concept of orality, of the possibility that singers can create and pass on to subsequent generations such highly complex verse without the aid of writing, is an idea so powerful that its full implications are only beginning to dawn on us; and in this sense I believe that Lord did well to hold firm to his original insights until the idea could sink in more fully. It is up to the present generation of scholars, Lord’s successors, to work out in detail the implications of “orality” and to modify the oral theory as it becomes necessary. Such work must be done with scholarly rigor, however. Neither impressionistic judgments derived from summaries and translations, nor the a priori conviction that poets must of necessity have writing in order to produce superior works of art, are acceptable. The job of scholars is not to pronounce something impossible simply because they cannot conceive of it. It is rather to set seriously about the conceiving of it.
Ronelle Alexander
Professor, Slavic Languages and Literatures
University of California
Berkeley, California

Hugh Lloyd-Jones replies:

Far from wishing to give the impression that the belief that Homeric epics were composed orally was on the verge of becoming extinct, I ended my review by remarking that it was still widespread and by mentioning a few of the learned scholars who still hold it; and Professor Riggsby’s well-informed, well-argued, and courteous defense of it is a further testimony to its continued vitality. The truth is that Milman Parry surely proved that the tradition of poetry to which the poems belonged had originally been oral and had doubtless continued to be so for a long period, so that they reveal many characteristics of oral composition which may easily be adduced as evidence that they were composed orally.

Poems composed orally can be preserved by human memory for very long periods, as the case of the Vedas is sufficient to attest. But what more than anything else persuades me that the Homeric poems cannot be the production of “simultaneous composition-in-performance” is the existence of subtle links between the different parts of the poems. To demonstrate this would unfortunately have required more space than my already long review could have commanded; the modern scholars who have done this best are Wolfgang Schadewadt, Karl Reinhardt, and Uvo Hölscher. For the understanding of Homer’s poetry, German is a more important language than Serbo-Croatian. Naturally, too, people who like myself think that the Greek alphabet had already been in existence long before Homer’s time are more inclined to believe that he used writing than adherents of the orthodox belief that the alphabet came into existence as late at the eighth century.

Those of Shive’s reviewers who condescended to pay serious attention to his arguments have indeed pointed out some weaknesses. But even if one entertains the reservations expressed by Professor Riggsby, one must agree with W.C. Scott (American Journal of Philology Vol. 110, 1989, p. 346) that “no reader can deny that Shive has demonstrated difficulties in Parry’s rigidly stated theory, and that he will make later followers of the master question some of their statements about ‘economy’ and ‘extension.”‘

Professor Alexander justly points out that my chosen examples of Serbo-Croatian poetry do not suffice to give an adequate notion of its character, and justly reminds us that a reader ignorant of a poem’s language cannot expect his judgment of its quality to count for much. As I remarked, “the shorter Serbo-Croatian poems are less important for comparison than are the two epics dictated to Parry by Avdo Mededovic, one of which has over 12,000 and the other over 13,000 lines.” I regret my inability to read these in the original, but would still remark that the summaries and translated extracts seem to me to indicate that they are epic poetry simpler and less profound than that of Homer, and therefore of limited value for the understanding of his art. I cannot share Professor Alexander’s feeling that the late Albert Lord “did well to hold firm to his original insights until the idea could sink in more fully.” The “insights” were Parry’s; the idea has now had plenty of time to sink in; and even some believers in the theory, not to speak of its critics, have put forward modifications that Lord might have been wiser to take note of.

To the Editors:

I completely agree with Professor Lloyd-Jones’s argument that the Homeric epics are highly sophisticated written works and I am impressed by his masterly summary of the complicated twentieth century debates around them. However, when describing my work he writes: “Bernal points out with much reason that it is likelier that letters were added than that letters were lost, so that the alphabets of Crete, Thera, and Melos, which lack the supplementals, may have been earlier not later, than the rest.” This is in fact the conventional wisdom and is diametrically opposed to my own position. I believe the Greek alphabets lost letters rather than gained them and I argue that the short alphabets of these Dorian islands developed later than the longer Ionian ones containing Φ, ψ and ω. After considerable study of the modern evidence on the subject, I have found myself back at the position held by Herodotus. When describing inscriptions that he believed to have been engraved before the Trojan War, he wrote: “In the temple of Ismenian Apollo at Thebes, in Boeotia, I have myself seen cauldrons with inscriptions cut on them in Cadmean letters most of them not very different from the Ionian.”

Martin Bernal

Cornell University
Ithaca, New York

Hugh Lloyd-Jones replies:

I must apologize to Professor Bernal for carelessly misrepresenting his position regarding the supplemental letters. But I myself adhere to what he calls the conventional wisdom in this matter. Since writing my review, I have become acquainted with an important new article by Joseph Naveh called “Semitic Epigraphy and the Antiquity of the Greek Alphabet” (Kadmos 30, 1991, pp. 143–152), which in my view strengthens his earlier argument that the Greek alphabet was invented as early as around 1100 BC. Naveh writes (p. 148), “Being the first adapters of a new script, the Cretans and Therans (like the Hatrans about a millennium later) probably looked upon their script as part of their tradition, and hence reverently preserved the typologically earliest letter-forms. However, whenever the script was taken over by another Greek city, its new users had no such reservations regarding the introduction of various changes.”

Herodotus’ belief that inscriptions he saw in the temple of Ismenian Apollo dated from the time of the Trojan war should surely be taken with a grain of salt. Before Herodotus, no author says that Cadmus came from Phoenicia.

This Issue

May 14, 1992