Bryn Mawr Classical Review 97.3.21

Gregory Nagy, Poetry as Performance, Homer and Beyond. Cambridge: Cambridge University Press, 1996. Pp. 254. ISBN 0-521-55135-8.

Reviewed by Barry B Powell, Classics, University of Wisconsin-Madison,,

G. Nagy's latest book, which began as the J. H. Gray lectures presented in Cambridge, England in 1993, contains valuable insights and summaries of scholarship, yet little with which I can myself agree. One will not learn how poetry was performed in archaic Greece, or what is meant by "Homer and beyond." How, why, when, where, and by whom poetry was performed are historical questions unfriendly to the etymologizing and comparative methods which N. prefers. Despite N's repeated claims to work within traditions of the Parry-Lord theory of oral composition, he denies the theory of the dictated text, a keystone in the Parry-Lord model, and he fatally denies an essential difference between the singer who composed in performance (the aoidos) and the reciter (the rhapsoidos), who memorized a written text for public reperformance.

My own bias is plain: I accept the familiar Parry-Lord model as plausible and see no reason, through N.'s arguments, to abandon it. When you wish to offer controversial views, you must provide hard, sensible fact, and this N. does not do. There is much learning in this book, but in this review I want to focus on N.'s interpretation of the Parry-Lord model. According to Parry-Lord, Homer was an oral poet who could not read or write, a real man who composed in performance on the basis of inherited linguistic and narrative patterns. His poems were probably recorded by dictation: the poet himself did not enjoy the privileges of literacy, especially in the eighth century BC, the world he describes according to the historical and archaeological record. N., however, thinks that Homer is a culture hero, a paradigm, a symbol for the oral tradition, whose mutability is reflected in the textual tradition, which contains variants. The poems of Homer (if he existed) were not taken down by dictation, but somehow were "crystallized" over a long period, especially during the sixth century BC in Athens. In this way, through "crystallization," there came into being the text we have today. Those are the big issues, as I see them, and it is here I wish to register my complaint.

Chapter I, "The Homeric Nightingale and the Poetics of Variation in the Art of a Troubadour," begins by quoting Odyssey 19. 518-523, where Penelope compares herself to a nightingale, the daughter of Pandareos, who killed her own child Itylos. A variant reading of unclear meaning, poludeukea for the vulgate poluekhea "with many resoundings" to describe the nightingale's voice, suggests to N. that we have found two "original" readings, each depending on the mutability of oral poetry and neither, therefore, more correct than the other. N. takes the obscure variation as paradigmatic of mouvance, a term he borrows from the criticism of French medieval poetry where similar variant readings appear. Implicit in N.'s long discussion is that written versions of Homer, and of other oral poets, reflect oral performance. Because we find variation in oral song, and because we find variation in the textual tradition, therefore the textual tradition reflects the variation so common in oral song and not the errors of copyists or the like. Does N. therefore think that the Iliad and the Odyssey were sung by Homer, not taken down in writing, then sung by a successor nearly verbatim (except for such minor variations as poludeukea/poluekhea), still not written down, then sung by someone else, with still more mouvance and a shifting of lines here and there, new particles creep in, then in the sixth century BC sort of written down, and then in the fifth century BC really written down, but still mouvance going on, until the Alexandrians at last established our text? Yes, N. does believe this. But the Iliad and the Odyssey that we possess are not oral poems; they are texts based on oral performance, which is not the same thing. The monumental labor and expense required to record the Iliad and Odyssey ensure for most Homerists that the poems were recorded a single time, that there was an original text. What is the problem with such an assumption? We do need to explain the small variations found in Homeric manuscripts, as we must explain those found in any text, ancient and modern. Still, from the moment that the Homeric poems were written down, they existed as texts and were subject to the vicissitudes of any text created in any fashion. Here is a cardinal element of the Parry-Lord thesis: oral poetry composed in performance is always something new, and there is no fixed text; but a written text is a fixed text. A written text is no longer oral poetry, nor subject to the rules that govern the generation of oral poetry, although it began as such. The second chapter, "Mimesis, Models of Singers, and the Meaning of a Homeric Epithet," returns to the nightingale, whose song is for N. somehow symbolic of how oral poetry works (a pretty nightingale from Pompeii adorns the book's handsome cover). N. notes that "the melodic patterning of the nightingale's song, as opposed to that of the lark, for example, involves the operation of a paradigmatic or 'vertical' axis of selection, not only a syntagmatic or 'horizontal' axis of combination." (p. 40) In support of his position N. actually cites ornithological studies of phrasings in the nightingale's song. But what, really, does this have to do with how archaic Greek poetry was invented or performed?

Back to poludeukes/poluekhes, nine pages of complex analysis suggest that the mysterious poludeukes means "patterning in many different ways (or 'many times')," that is, the obscure variant actually describes the oral tradition itself. Poludeukes is also the name of one of the Dioscuri, Polydeuces, who is the Morning Star, who returns again and again, like oral poetry (though never exactly the same, perhaps). Twins imply at once sameness and difference (like oral poetry), and so the essential nature of the ever-returning and ever-varying "performance tradition" turns out to be encoded in Aelian's otherwise unattested poludeukea.

Now N. turns to mimesis. All song and poetry is mimesis, impersonation or reenactment, but for N. in the special sense that when a poet later than Homer performs the Iliad, he pretends to be Homer himself. The topic of a poet's attitude to his material is always interesting, but if all song is mimesis in N.'s special sense, whom did Homer pretend to be? By mimesis Aristotle seems to have meant something like fictional representation and enactment of experience, but N. speaks as if his own usage is somehow Aristotelian. It's easy to get lost in all this.

Chapter 3, "Mimesis of Homer and beyond," pushes forward the theme broached, then gets caught on the wire of the word "rhapsode." Denying an essential distinction between aoidos, Homer's own word for singer and what most critics would naturally call an oral poet, and rhapsoidos, Plato's word for the performer Ion and his ilk, N. seems to claim that aoidic and rhapsodic performances were really the same (though aoidoi came earlier); this would have to be true, if textual variants do reflect composition in performance, as N. believes. According to N., you can somehow go from unfixed to fixed within the oral tradition -- so the rhapsode does not learn from a written text, and he is still in some sense an oral poet, and there is in any event a continuum between the creative aoidos and the mimetic rhapsode.

N. seeks support for his argument from his derivation of the word rhapsoidos, and he has good company, ancient and modern, in understanding it to derive from rhapto, "sew," and aoide, "song," so that a rhapsode was a "song-stitcher." Passages in a Hesiodic fragment and Pindar do play with the image of stitching together song and may well reflect knowledge of the word rhapsoidos, not however attested until Herodotus, but the Greeks of course always enjoyed false, often playful etymologies (Homer derives Odysseus comes from odussomai "to be angry," Od. 19.407). We should depend on our own wits in understanding the origin of rhapsoidos.

From Plato's description of Ion, a rhapsode seems to have been someone who memorized a written text of Homer, then delivered the memorized lines at public festivals. Such rhapsodes as Ion were able and eager to comment on the meaning and importance of the text they mastered, and their dramatic delivery won them prizes and prestige. Ion specialized in Homer and was proud of it. An aoidos, by contrast, to judge from Homer's descriptions especially in Odyssey 8, entertained at private banquets or on the athletic field, composing his song to the accompaniment of the lyre, sometimes according to suggestions taken from the audience. Homer's aoidoi Phemius and Demodocus look very much like the guslars of Parry and Lord and according to the Parry-Lord theory they are like them. They don't look like Ion, and they are in fact not like Ion.

According to this view, which most Homerists accept, an aoidos was an oral poet, a guslar, who made a new song every time he sang, depending of course on traditional stories, traditional themes, and traditional language, but always adjusting his story to his audience and the needs of the moment. Oral poets (like Phemius and Demodocus) are not literate, according to the Parry-Lord model, or if they become literate, they do not depend on writing for the creation or reperformance of their song. The differences between an oral performance of the same song by the same singer can be great and are not confined to several lines here or there or single words, let alone particles. The stringed instrument that accompanies the oral poet's composition in performance, like the lyre of Homer's aoidoi, is essential to the song's creation and its melodies and figures are a conspicuous part of the song to anyone in the audience. Demodocus can play dance tunes unaccompanied by song; he is a highly skilled musician. All the music is lost when an oral song is taken down in writing; the written version is a skeleton, a snapshot of the language of the original song, for those clever enough to decipher it from a continuous stream of unaccented signs, without capitalization or word division, which takes training and a lot of practice. The deciphered, recreated version may be bereft of the music and social context of an oral poem, but new opportunities arise through the histrionic presentation of familiar passages.

A rhapsode was skilled in just such presentation, a kind of actor (N. himself says this later in the book), who prides himself on sticking close to the written text, the source of his authority, while of course making clear his own brilliant understanding of the text. Actors, and rhapsodes, surely reproduce inexactly the written text on which they depend, or alter it, but no such alterations matter unless they somehow are entered into a manuscript. Rhapsodes are literate and can make such changes if they choose; some of the variations we find in the Homeric texts, such as even poludeukea/poluekhea, may go back to rhapsodic alterations or errors or "improvements" (still, why would a rhapsode go to the trouble to make such inconsequential changes?).

Rhapsodes declaimed at the Athenian Panathenaea, dressed in elaborate costumes and holding a staff, a rhabdos, whence their name must derive. Ion is no musician, for the rhapsoidos does not compose in performance to the accompaniment of the lyre: he holds a staff as he declaims, so is a "staff-singer," as illustrated on Athenian pottery, as the auloidos was the "flute-singer" and the kitharoidos the "lyre-singer," each named after the item that accompanied his presentation; auloidoi and kitharoidoi performed side by side with rhapsoidoi at the Panathenaea and other Greek festivals. "It is simplistic and even misleading to contrast, as many have done, the 'creative' aoidos ['singer'] with the 'reduplicating' rhapsoidos," N. writes (p. 113), quoting from an earlier publication. But it is not simplistic, nor misleading, but critical to our understanding of archaic Greek literature to recognize that aoidoi and rhapsoidoi were different kinds of men, who did completely different things in the world. Above all, rhapsodes could read and write and aoidoi could not. If the Greeks had not themselves given us words by which to distinguish oral poets from literate representers of oral poetry imprisoned in alphabetic writing, modern historians would have to invent such words. Why, therefore, insist that they are the same thing?

Building on a false etymology for rhapsoidos as "song-stitcher," the remainder of Chapter explores imaginatively the theme of stitching and sewing. Discussing a passage from Plutarch about young girls with chitons not sewn at the bottom, N. writes: "Just exactly where you sew together -- and leave off sewing together -- becomes an exquisite art of tailoring to suit the senses and sensibilities of the viewer." (p. 69) But Plutarch wants to say how Spartan women are forward and immodest, unlike Roman women; there is no art here, but morality. Discussing Achilles' singing klea andron in Iliad 9 (p. 73), N. explains that Patroclus is waiting his turn to sing after him, just as rhapsodes did at the Panathenaea. Did, then, Achilles and Patroclus have a text of songs, imported from Chios, whose matter they were required to "go through in sequence by relay," a rule Hipparchus set down for rhapsodes at the Panathenaea? For Parry and Lord, Homer was a guslar, but for N. Homeros is "he who joins together," his name derived fantastically from homo and ar-, whence "I propose, further, that there is a corresponding parallelism between the concept of Homeros and rhapsoidos. Homer is no longer Homer, but a culture hero, the font of song." (p. 60) Most will find this hard to believe.

Anthropology has a good deal to offer literary studies, and in Chapter 4, "Mimesis in Lyric: Sappho's Aphrodite and the Changing Woman of the Apache," N. turns to an anthropological study of a Navajo girl's puberty ritual, where the girl is identified with Changing Woman, a figure in Navajo religion. Magical chants encourage the identification and the girl is matched with an elder attendant whom she ever after calls mother. N. compares the relationship between the girl at puberty and the elder attendant with that between nurses and heroines in Greek tragedy. But the Navajo ceremony is a ritual, not a drama, and the Navajo ceremony does not depend on a written text; the Navajo initiation ritual has no author. Euripides, by contrast, wrote the Hippolytus, his actors memorized and performed it, and the nurse to whom he refers was a familiar figure in Athenian life.

N. also suggests that the identification between Changing Woman and the Apache girl undergoing initiation is the same as that between the "I" in archaic Greek lyric and choral verse, of which so much has been written: "I must insist that this kind of 'acting' in the context of archaic Greek poetry is not a matter of pretending: it is rather a merger of the performer's identity with an identity patterned on an archetype -- a merger repeated every time the ritual occasion recurs." (p. 96-97) Here again is N's theory of mimesis. But archaic Greek lyric was not ritual either, and performers of it certainly did pretend, as when a male symposiast referred to himself as "I, Sappho." And what, exactly, is an archetype? At what level can a Navajo ritual initiation, accompanied by magical chant, be the same, parallel to, or explicative of poems composed more than 2,500 years ago in writing by strong personalities in ancient Greece, to be reperformed by others, most often within the drunken all-male symposium? Ceremonies that initiate into puberty are found everywhere in the world, but only in Greece did archaic lyric appear, with its mystery and glamour. N. has not proven the usefulness of his example.

Chapter 5, "Multiform Epic and Aristarchus' Quest for the Real Homer," sets down what N. calls the Five Ages of Homer, a summing up of the various stages of textual fixation according to his model of the evolution of the Homeric poems. First was a fluid period, roughly 2000-800 BC; second, a "pan-Hellenic" period, still no written text, down to the middle of the sixth century; third, "potential texts in the form of transcripts," from mid-sixth to fourth centuries; fourth, a standardizing period with "texts in the sense of transcripts or even scripts" under the supervision of Demetrius of Phalerum from 317-307; finally, a rigid period "with texts as scripture," from middle of the second century BC on, beginning with Aristarchus' work around 150 BC: "In brief, then, this scheme of five periods in Homeric transmission brings into play primarily the dimension of performance, in particular the traditions of the rhapsoidoi, and, secondarily, the dimension of text as a derivative of performance, where each successive period reflects a progressively narrower concept of textuality, from transcript to script to 'scripture.'" (p. 113) The whole process N. describes as "crystallization," a word used in his earlier publications.

Here, then, is the heart of N's thesis, and of his dilemma. For a written text is not the product of a chemical reaction (and what is a "potential" text? What is the picture exactly?). In archaic Greece texts seem to have come into being in three ways. Scribes took down verses from oral poets by dictation. Why does N. not accept A. B. Lord's theory of the dictated text? N. never explains why. Aristocrats, who learned how to decipher texts that began as oral poetry, discovered how to create, in writing, new forms of poetry, lyric and choral song (how do you train a chorus without a written text?); Sappho and Euripides composed for reperformance from a written prompt in just this way. Finally, texts which originated in either way, through dictation taken from oral poets or created originally in writing, were copied when one man read aloud from a written text and another wrote down what he heard, or he copied the manuscript by eye (we're not very clear about this). Texts may be a derivative of performance, as N. puts it, when an oral poet dictates to a scribe (which N. denies happened), but when a rhapsode or actor delivers orally a memorized text, performance is derivative of text. The model seems so persuasive that I do not see why N. is determined to overthrow it. We have people dictating texts, creating texts in writing, and copying texts. "Crystallization" is an unfortunate metaphor and I cannot see why N. wants it.

The balance of chapter 5 contains an interesting discussion of personalities and dates at the Alexandrian Mouseion, also plausible explanations of how the Alexandrians got their texts and which ones. N. is especially interested in the role that Demetrius of Phalerum may have played in the transmission of the Homeric texts, a theory he explores further in the last two chapters, "Homer as Script" and "Homer as Scripture." He may be right. Both chapters also contain valuable quotations from Athenaeus and other ancient sources which cast light on the problem, What was a rhapsode? (but by no means affirm N.'s conviction that rhapsodes were latter-day aoidoi). I profited, too, from N.'s discussion of Aristotle's contribution to the textual tradition and of what is meant by the Homeric koine, the heart of the Homeric Question. Here N. is best, when he begins to approach the evidence historically.

The book is handsomely produced, with few errors (but Nemean 2.1 is identified as 1.2 on p. 63; three inaccurate accents on page 111, and on page 170 a line has dropped out and another repeated on page 171).

N.'s book is not easy to read. An opaque style sometimes obscured by jargon harms his argument throughout, as when discussing Aristotle's theory of mimesis he writes: "So long as the represented 'that' remains absolute, that is, absolutized by the myth -- the representing 'this' remains a re-enacting 'this.' So long as 'this' imitates an absolute 'that,' it re-enacts as it imitates; the re-enactment remains primary and the imitation remains secondary." (p. 55) N. also has the annoying habit of italicizing words or placing them in parentheses, as if by emphasis to replace precise expression, and an oracular posture sometimes replaces argument with rhetoric ("I now propose that Demetrius legitimated the evolution of a corps of Homeric performers who were relatively more professionalized than earlier performers." (p. 174) The argument itself never gets off the ground because N. lacks a good model of the relationship between written text and oral poem, or a clear picture of how written texts came into being. He seems unaware of the special role that alphabetic writing, as a technology, played in the preservation, creation, and dissemination of archaic Greek poetry. N. does not accept that an oral poem ceases to be an oral poem when it becomes a text. Without saying why, he rejects Lord's theory of the dictated text, a compelling explanation of how the Iliad and Odyssey came into being, to offer a second model so vague as to send us scurrying back to Lord. Originality cannot be won in the face of evidence. Nowhere, in the vast field of song-collecting in modern times throughout the world, do any parallels exist which might fit N.'s proposal, whereas Lord's experience has been repeated over and over again. I regret writing so querulous a review, especially about an author who has given us much to think about over the years, but in my opinion N.'s views, received uncritically, obscure difficult truths hard won about Homer, the divine Homer, and his inimitable poems. There, in the end, must our loyalties lie.