Chapter 3. How Oral is Oral Composition?

In the two preceding chapters the main interest was the question of the formula, which we approached from a number of perspectives. The present chapter will start widening the horizons by asking the question that ought perhaps to be at the center of the study of oral poetry: the question of orality itself. What does it mean for a poem or a poetic tradition to be “oral”? What is “oral style”? And, finally, the most basic question: what exactly does “oral” mean? I start from this last question, and hope to deal with it in such a way that answers to the earlier ones suggest themselves in due course.

We usually see “oral” in opposition to “literate,” “orality” to “literacy.” We sometimes speak of “oral” versus “literate,” implying that somehow the two terms exclude each other. But sometimes we seem to assume that the two may coexist. So it seems that our usage of the very term “oral” and its abstract “orality” needs to be specified. What exactly does “oral” mean and in what senses do we use the term? Semantic clarity on our own usage in our own culture can only be beneficial when it comes to the study of Homeric orality.

A first sense naturally presents itself in the observation that a discourse may be either read or listened to. In the latter case, we can say that it is delivered orally. The term “oral” here applies to a possible way in which we can use language: by speaking. A more precise term might be simply “spoken,” as a name for one possible medium for our discourse. [1] In this sense of “spoken,” we can oppose “oral” to “written,” in order to bring out the seemingly straightforward difference between the human voice on the one hand and the graphic display of characters on the written or printed page on the other.

The situation becomes more complicated when we turn to a second sense of “oral.” Speaking and writing, sounds heard and visual signs seen, are different media for using language, but the difference goes deeper than the obvious difference between sound and sight, or hearing and reading. Speaking and writing for us are different activities that call for different strategies in the presentation of a discourse and its comprehension by its recipients. When we address people in an actual speech situation, even a formal speech situation, we obviously use our language differently from when we address our absent readers in our books and articles. So either medium in our culture, speech and writing, comes with its own set of associations, even its own mentality. To bring out this aspect, we might speak of “oral” as the conception that underlies a discourse, and oppose this quality not to “written” but to “literate.” [2]

This sense of “oral” has to be distinguished from the sense related to medium. To begin with, we note that both terms may not apply at the same time. A discourse may be oral as to its conception, but written as to its medium: it has been written down, but does not display the features that we normally associate with a coherent written text. [3] Conversely, a discourse may be written, i.e. literate, as to its conception and oral (“spoken”) as to its medium. This often happens in the oral presentation of academic papers, when the author has not succeeded in hiding the “writtenness” of the lecture.

But the difference between “oral” as a medium and as a conception goes further than this. As a medium, “oral” excludes “written” in the sense that at the moment of its reception a discourse usually cannot, for us, be both spoken and written at the same time: it has to be either phonic or graphic. In the dimension of conception, on the other hand, the relation between “oral” and its opposite is quite different. “Oral” and “literate” can here be seen as the two poles or extremes of a continuum, with numerous gradations in between. On the one extreme, a discourse is maximally “oral,” making full, and necessary, use of the requirements of the spoken medium; on the other extreme a discourse is maximally literate, written exclusively according to the requirements of the written medium in the culture at hand. In practice, most discourses will display both oral and literate features in varying ratios, being situated at some point between the two extremes.

An important consequence of the understanding of “oral” in terms of a discourse’s conception is that the orality of a discourse is quite compatible with writing, and hence with reading. Equally important, especially for the study of how oral traditions come to be recorded, is that the very notions of writing and reading are sensitive to the new understanding of “oral.” Often we talk about reading and writing as monolithic concepts, and in the study of Homer as oral poetry this has led to confusion (see below); but in reality these concepts run the gamut of the whole oral-literate continuum. In other words, the conception of a text is indissolubly connected with the conception of its writing and its reading. The three continua, that of a discourse, its writing, and its reading, can be displayed graphically in the following way:

In the first continuum, we see “oral” and “literate” as extremes, end points defining the nature of the space in between. The second continuum specifies what happens when the written medium enters the picture; the continuum is the space between the writing of a discourse that is oral as to its conception, here termed transcription, and the writing of a discourse whose conception is unequivocally literate, a matter of written composition. “Composition” constitutes writing in our culture’s understanding of the concept. When we speak of writing a book, an article, a letter, or even a sentence, we usually mean that we are composing the text in question, the physical act of writing being taken for granted, and increasingly not occurring at all, as in the case of the present text, which was composed (and recomposed) by its author at the computer screen.

Toward the left-hand extreme on the other hand, the physical act of writing is essential to the conception. Writing here is to be understood not as composing a discourse, but as transcribing it. The act of writing transcodes a discourse into the opposite medium, and as a translation of sounds heard into signs seen it has little to do with the composition of the discourse in question. In other words, the composition of such a text takes place independently from its writing, and may well precede that act in time. [4] Reading at this point in the continuum, conversely, retranscodes signs seen into sounds heard: the reader’s voice turns sight into sound, and in doing so is just as physical as the writer’s hand. No such physical operation occurs at the literate end point of the reading continuum: “voice” here is fictional, metaphorical, and resounds in the reader’s mind, if at all, not in the airwaves. We may also see the scale, then, as one of increasing fictionalization: the further we move to the right, the less likely it is that any actual spoken discourse precedes or follows the text in time. Writing and reading in our modern sense of composition fictionalize the speech act, inventing a textual voice that in the written reception of the text has to function as the substitute of the absent author’s actual voice. [5]

The distinction between medium and conception, we are now in a position to see, is itself sensitive to the continuum constituted by “conception”: at the right extreme, where academic and scholarly writing occurs, reading and writing are typically achieved through the graphic medium of the written or printed text, without the opposite, phonic medium coming into play. But when we move leftward in the direction of the oral extreme of the continuum, we observe that the two media of sound and sight, voice and text, become more and more intertwined. Both reading and writing here take place through the crucial intervention of the human voice that serves as the substance to be encoded into text in the act of writing and as the decoding tool in the act of reading. No silent reading takes place here, and the discourse produced by the reader’s voice is a reenactment of the writer’s voice that was transcribed in the act of writing.

The three interrelated continua apply to various dimensions, of which the most important for our purposes here is the historical. Historically, the left-hand extreme of transcription can be seen as the moment when writing first enters a society, a moment that is different indeed from the point we have reached to date in the age of computers and the Internet. [6] The strangeness of routines that run counter to scholars’ conception of literacy and textuality becomes now part of a wider historical problem: what did it mean to read or write at the time of the earliest texts in archaic Greece? What consequences do such reading and writing have for the orality of the Homeric poems?

The preceding discussion allows us perhaps to get a better picture of the problems involved here. I have distinguished between two senses and understandings of the concept of “oral.” In the sense of “oral” as medium, whether we think of speech and writing, voice and text, as separate activities or as integrated in the act of reading and writing, neither of the two has a privileged position with respect to the other. In the dimension of a discourse’s conception, on the other hand, matters are more complicated, because here the opposition is far from neutral: an asymmetrical, hierarchical relation obtains between the two. We endow the terms involved, “oral” and “literate,” with a cultural value. So we speak of “oral poetry,” not to characterize a given poem as spoken (when it is read aloud), but to define it with respect to poetry as we know it in our culture. Similarly, we speak of “orality,” not to describe what happens when someone talks, but to label a period or a culture as different with respect to our own, literate culture. And most importantly, we tend to consider texts that are oral as to their conception as crude or primitive, simply because we use our own sense and conception of writing as a norm.

I believe that it is important to realize that not only the spoken medium of language, but also the spoken conception is a phenomenon that deserves to be studied in its own right and on its own terms, rather than from the point of view of writing. When we present a spoken discourse, with its oral conception, as text, we tend to think of that text as “primitive” or, in another context, as “archaic,” without realizing that when we talk ourselves, our own discourse, when transcribed and presented in written form, may look very similar. The discourse was never meant to be read in our understanding of that concept, and looks primitive only because we subject it to the reception conditions of the written medium. [7] Conversely, a discourse that is written as to its conception will in most cases be equally unsuccessful when it is presented as speech, because the rhetoric of its writtenness is not adjusted to an ordinary speech situation.

This leads us to the conclusion that the use of the terms “oral” and “orality” can often not be separated from our own literate perspective. Instead of constituting a neutral contrast with “literate” and “literacy,” they are “the other member” in a contrastive pair orality-literacy, in which we align ourselves with the literate member. In this sense “oral” and “orality” are bound up with literate culture. They denote a lack of the literate conception of language, and so define speech as the construction of a writing culture that uses its own absence to define its opposite: a metaphysics of writing. [8] As a further illustration, let us imagine a society in which writing, in whatever form or importance, is totally absent. In such a society poets may well exist, but in the absence of literate poets they cannot be oral poets. Nor have people an “oral style” in this society, because there is no literate style to compare their speech with. The perspectival bias inherent in the term “oral” is similar to that in “natural” or “organic” as applied to food in a society of hunters and gatherers. These terms presuppose their opposite, and are meaningful only in a situation in which the opposite is the norm.

The typical consequence of such literate bias is a neglect of speech when a discourse that is oral as to its conception is viewed in terms of writing (as in the case of archaic sepulchral or dedicatory inscriptions), and, conversely, a neglect of writing when it is viewed in terms of speech—this is the case, I submit, with the Homeric tradition, to which we now turn.

Let us start with the orality-hypothesis, the theory of oral composition proposed by Milman Parry and given an authoritative and influential form in Albert Lord’s The Singer of Tales. I do not wish to detract from the brilliance of Parry’s original insights, nor from the importance of The Singer of Tales in opening up vast areas of fruitful research, and in serving as a powerful heuristic guide in the study of traditional epic poetry. But I do believe we act in the spirit of Parry’s pioneering work when we test his findings against subsequent developments in the study of oral poetry and spoken language, and adjust or modify them when necessary. In particular, I would like to assess the notion of oral composition within the framework that I have tried to sketch.

The concept of oral composition has been central in the discussion of Homer in the wake of Parry’s and Lord’s arguments. Lord thought of oral and written composition as mutually exclusive:

And again:

For Lord, oral composition was the ultimate criterion on which the orality of oral poetry depended, and he transferred the results of his fieldwork on the South-Slavic guslari to Homer. In order for Homer to be oral poetry, the Iliad and the Odyssey had to be orally composed. If one doubts the oral hypothesis or tries to weaken it, one loses Homer as oral poetry. This inflexible position was reflected in the subsequent scholarly discussion, where we could see Homerists divided into two segregated camps, the oralists and the “scripsists,” [10] with oralists coming sometimes close to minimizing and trivializing the role of the Homeric text, and scripsists assuming that the mere existence of that text is a green light for unreflective literate and literary study of Homer.

I believe that both positions are anachronistic and ill-taken, the oralist position in principle no less than the scripsist position. The above discussion of the two senses of “oral” might help us see how both positions ask the wrong questions and seek the wrong answers. In fact, Lord’s very notion of “oral composition” amounts to a conflation of the two senses discussed. The idea of oral and literate composition as mutually exclusive is in line with the understanding of orality as medium, in which a discourse, in our culture, is either spoken or written, not both. But we saw that “oral” as a conception is a gradient property that is not at all incompatible with writing, provided we are prepared to accept that our own conception of writing is culture-dependent. Relativism as regards writing, however, is not to be found in the work of Lord and other early oralists, who have consistently assumed that writing inevitably kills off an oral tradition and blunts the capabilities of oral poets. [11] The anachronism of such a position is obvious in light of the growing historical awareness as regards writing and written communication: we begin to realize that our own notion of writing is in need of anthropological or historicist correction when it comes to the study of writing in other times and places, such as the Greek situation in the seventh or sixth centuries BCE. If we use the Middle Ages as comparative historical evidence, we notice that such things as “writing” and “reading” in cultures other than our own can be surprisingly “oral,” in the sense that the human voice and ear play an important role. [12] Archaic Greece can contribute material to the discussion through Jesper Svenbro’s reading of archaic funereal inscriptions. [13]

It does not seem impossible, then, that writing played a role in the composition of the Iliad and the Odyssey. A writing that is far removed from the written composition that we routinely have opposed to oral composition. Even though the Homeric text as it was first written—a text that is surely very different from our received text [14] —cannot have been an exact transcript of an oral performance, its conception was, and remained, profoundly oral. And the medium of Homeric poetry remained oral too, for the simple reason that performance, and not silent reading, remained the only medium of presentation of Homeric poetry till well into the classical period. [15]

But this position as regards Homer and writing does not imply that the Homeric text should be seen as “post-oral,” as is often done, or as transitional between orality and literacy, or some such qualification. To see Homeric poetry as somewhere “in between” orality and literacy would still be to view orality and literacy as two atomic, mutually exclusive opposites. The point is rather that writing in the Greek archaic period must have been so different from our notion of writing, so “oral,” in fact, that the simple dichotomy between orality and literacy breaks down.

This leads me to a further characteristic feature of the position of Lord that becomes apparent in light of the discussion on perspectives earlier in this chapter: the notion of oral composition is an unmistakably literate notion. Oral composition presupposes an oral style as opposed to a literate style, but, as we already saw, the very notion of “oral style” betrays the perspectival bias of the literate classicist. “Oral style” presupposes “literate style” in the same way that “archaic style” presupposes “classical style,” or “paratactic style” presupposes “hypotactic style,” and so on: the oral nature of this style is of our own making, and defined with respect to literate style in the black-and-white contrast that I have already argued against.

The basic ingredient of this oral style, and according to Parry and Lord the necessary condition for oral composition, was of course the formula. The formula was defined as a kind of repetition that was metrically useful, indeed indispensable, in the oral composition of the epic story. Again the binary contrast with written style is apparent: written style does not display this kind of repetition, because the composition of the written text is not dependent on it. In its nonformulaicness written style functions as perspective, even as norm, in the definition of oral style, and of the formula as its prime criterion. Language, we wrongly assume, is non-repetitive and “original” by default.

We may also note—and this will become important later on in this chapter—that the same principle, but now in reversed form, applied to enjambment. Whereas the formula could be seen as a prime feature of the oral style (so that there should not be too few formulas in a text if it is to be oral), enjambment was seen as a feature of written style, so that there should not be too many cases of enjambment in a style that was supposed to be oral. Accordingly, the relative frequency of these two features has been a major issue in oral-formulaic theory. [16]

At this point I may have to forestall a possible misunderstanding. The contention is not that Homer does not have formulas. Rather, I am suggesting that we have to be aware of the perspective involved in defining them. The presence of formulas in Homer is of course undeniable, and so is the metrical function that Parry assigns to them (though I would give priority to their meaning as recurrent speech rituals in the performance). [17] But more important for my present purpose is the direction from which they are approached: not from the point of view of literate style, but from the point of view of speech.

I have already argued that we have to be careful in imposing our own culturally dependent conception of writing and written style on texts and discourses that are different. This means that we may have to pay attention to the oral conception of language for its own sake. In other words, we have to ask what it means for language to be spoken. This question is, I think, of the greatest importance question for the study of oral traditions, whether they are written (textualized) or not. And when they turned themselves into text, and we have to study them as text, we need to do some thinking before we apply the concepts and assumptions deriving from our own written conception of language.

An awareness of the oral conception of language and its specific features may lead us to a study of Homeric poetry as speech, regardless of when or how the poems were committed to writing. That is, it may be worthwhile to look at the poems as spoken, not written language, and untouched by the preconceptions of written style and composition. Such an analysis does not aim at a simple equation of Homeric discourse and ordinary speech. But the most conspicuous features of Homeric style, such as meter, formulas, and parataxis, are amenable to the analytic tools and concepts used for the study of speech in general. One might speak of “special speech” in an effort to replace “oral poetry” with terminology less reflecting the perspective of the literate researcher.

Special speech in an anthropological perspective is the discourse in which a community asserts its identity in establishing links with a meaningful, heroic past, or with a spiritual au-delà. [18] Such discourse is obviously different from what ordinary people say in ordinary situations, yet it will draw for its effect on whatever feature of the language in question is most prominent. In the case of the Homeric tradition, I argue, this stylization of ordinary speech results in meter and formulas as the prime feature of the discourse of the epic performance. [19]

Formulas, the basic units of special speech, seen as the stylization of ordinary speech reveal an instructive aspect of the confrontation of the written with the oral conception of language. Of what are formulas the stylization? The basic unit of language is usually thought of as the sentence. Yet this concept, and especially our understanding of it, is a matter of the written conception of language. By this I do not mean that sentences are confined to discourse with a written conception, and that speech does not have them. Manifestly, in spoken language, sentences occur too, but the difference from written language is that in speech the sentence is not a stylistic norm, a criterion for acceptability. In other words, a written discourse in our sense is flawed when it does not attain acceptable sentential syntax; a spoken discourse, on the other hand, may still work when its speaker does not arrive at sentential structure. In fact, such a structure may well not be in the speaker’s interests at all: it may be too complex to grasp by someone who has to listen on the spot, and who cannot review the construction in its entirety on the printed page. Such a discourse might look awkward in writing, but, as described earlier, this may have more to do with mismatch of two media (the one medium subjected to the reception conditions of the other) than with any lack of “quality” of the discourse in question.

Instead of the sentence there appears to be another unit that is of more direct relevance for the study of spoken discourse in any language. This is what has been called in linguistics the “idea unit,” or “tone group,” or “intonation unit,” or some such characterization. This kind of unit is usually four to seven words long; it can be a complete syntactic unit, such as a clause with a verb and an object, but it can also be something that needs to be complemented to make sense syntactically; and in spoken language it is marked by intonational boundaries, and often by pauses. The linguist who has most widely written on the intonation unit is Wallace Chafe. [20] Chafe adds a psychological, cognitive dimension to this feature of spoken language, by linking the intonation unit to such human faculties as consciousness, memory, and perception. An intonation unit, as Chafe argues, is the verbalization of the amount of information that can be held in what cognitive psychologists call the working memory, or short term memory, or what Chafe himself calls the focus of consciousness. In this way the intonation unit becomes the basic element of speech as a flow through time. Both the speaker and the listener have to produce and process the discourse on the spot in a temporal flow in which returning or looking back is impossible. In this way they are left to the resources and limitations of their cognitive makeup.

Now the intonation unit is not only a universal manifestation in speech of the workings of human consciousness; it is also the point of departure for any enhancement of speech in the form of oral traditions, genres of special speech. The intonational and prosodic properties of the unit can be stylized into metrical properties; and their cognitively determined length makes them the ideal basis for formulas as the basic ingredients of epic discourse. In other words, the intonation units of ordinary speech become the metrical units of special, poetic speech. [21] I propose, then, that it is the intonation unit of spoken language that lies at the basis of essential aspects of Homeric style, such as the very concept of formula, as well as of parataxis and colometry. As an example, let us study the passage describing the first appearance of Andromakhe in the Iliad:

The text of Monro and Allen (whose punctuation I have followed in the Greek column) presents this passage as three sentences, one comprising my units 1–4, the second one 5–19, and the third one 20–28. This punctuation is arbitrary, however, since sentence breaks are equally possible after units 1, 9, 20, or 25. Moreover, the break after 19, the end of the second printed sentence, is unclear, and depends on whether we take ἥ in unit 20 as a relative or as a demonstrative pronoun. But such a grammatical distinction is one of our own making and is often irrelevant for Homeric discourse (notice also units 8 and 18 in this regard); here it serves as a questionable basis for punctuation. Although the full stops we print in our Homer texts may seem unquestionable and felicitous in many other cases, they remain attempts at accommodating Homeric discourse to the requirements of the written medium in our conception.

Of more direct relevance than sentences for the analysis of Homeric poetry are the units into which the passage naturally divides. Most of these units are one half-line long, in length and duration comparable to the intonation units that are observable in ordinary speech. To be sure, this speech analysis does not mean that I see the passage as improvised, or orally composed in the sense of Parry and Lord. We will never know how close the passage is to the textless performance of the singer of tales—perhaps less close than we would like it to be. All we know is that the text of the passage is characterized by a thoroughly oral conception, and so far removed from our conception of a written text.

Instead of sentential arrangements, we see a relation of addition between units: each unit builds on the other in an ongoing flow of incoming detail. [22] Straightforward examples of such adding units include prepositional phrases (unit 4), locative expressions (16), participial phrases (17), proper names (12), and qualifying adjectives (23, 24, 25). This appositional nature of Homeric syntax has been often observed, in particular by Parry’s teacher, Antoine Meillet. Meillet, the Indo-Europeanist, observed that Homeric Greek, like other older Indo-European languages, is characterized by a pervasive tendency for words and phrases to be autonomous, not governed by other constituents in the sentence. [23] Meillet opposes this relation of “apposition” of Homeric Greek and Indo-European to the more hierarchical relation of “rection” of modern languages such as English or French, where linguistic elements tend to be governed by one another. [24]

It is certainly true that there is a syntactic, typological difference between English or French on the one hand and Ancient Greek on the other; but the confrontation of the written with the oral conception of language plays a role too. [25] To take a simple example, in the sentence Peter gave the book to John, there are three nouns (“Peter,” “the book,” and “John”) that are governed and held together by the verb “gave.” In Indo-European, according to Meillet’s reasoning, such noun phrases would be much more independent, and we would get such configurations as Peter, he gave it to John, the book or The book, Peter gave it to John. But a sentence like Peter gave the book to John, in which one single verb governs no less than three substantives, would be very unlikely to occur in spoken English or French, too, as it is produced by natural speakers in natural situations. [26] The information it contains would be preferably broken down into two or more intonation units that resemble Meillet’s autonomous phrases. It seems natural, then, to assume that speech syntax vis-à-vis Meillet’s own literate conception of language has played a role in the formulation of the Indo-European appositional construction.

We can now consider units 10–11 and 18–19 in the passage above in the same way:

Units 11 and 19 are complete clauses; their verbs ἦλθε and ἔχεθ᾿ do not need ἄλοχος and θυγάτηρ in the preceding units to be syntactically complete. So under this analysis, the nominative substantives in units 10 and 18 are not the subject of their verb. In Meillet’s terms, these substantives are autonomous elements related by way of syntactic agreement to the subsequent discourse; Devine and Stephens would speak of “null anaphora” and “adjunct lexical argument.” [27] For the purpose of the present argument, the substantives are part of a separate unit, filling its own half of the verse, and preceding its clause.

Terms such as “addition,” “apposition,” or “parataxis” might seem to imply that the progression of Homeric discourse is random and unable to convey hierarchical relationships between the units of discourse. But this is not a valid inference, unless the terms themselves cannot but convey the absence of literate stylistic virtues. By the notion of a flow of discourse, in which the speaker leaps from idea to idea, I do not mean to imply that there is no planned organization. A spoken discourse may be different from a written discourse in its structure and organization, but it is organized all the same. An important aspect of it can be captured by the notion of goal or preview. [28] In spoken discourse, something can be said not as information in its own right, but with an eye on a situation to be reached in due course: a detail may be stated in order to be explained, which will lead the listener to a goal that was indicated earlier. Such a preliminary indication occurs at unit 13 in our passage: the identification of Andromakhe as “the daughter of Eëtion.” Unit 14, which takes up the name of Eëtion as starting point for information about his dwelling place, is not a digression which leads the narrative off the track. Eëtion’s identity is crucial for that of Andromakhe, [29] and instead of being a piece of gratuitous detail, unit 13 looks ahead at 18, the moment when Andromakhe returns in the discourse, now fully contextualized and identified. This moment is formally marked as previewed by the repetition of θυγάτηρ ‘daughter’, just as ἤντησ᾿ ‘came to meet’ two units later (20) takes up ἐναντίη ἦλθε ‘came to meet’ (unit k) to mark the final transition from Andromakhe’s relevant identity to her actual appearance on the scene. This is ring composition, of course, but I would stress that this phenomenon is less a feature of “oral style” or “archaic style” than a naturally occurring phenomenon in the presentation of speech. [30]

Since Homeric speech units are stylized speech units, it is impossible to ignore their metrical dimension, in particular their relation to the verse as a rhythmical unity. This leads us naturally to the notion of enjambment. This term surely denotes an important feature of Homeric style and poetry, but to the extent that “sentence” figures in its definition (enjambment being the mismatch of sentence and verse) we have to be prepared, again, to acknowledge that our own literate conception of language and texts may be involved. I propose, then, to view enjambment in terms of the basic units that make up Homeric discourse, rather than in terms of the sentences that we impose on it.

Let us first return to the appearance of Andromakhe. For most of this passage, units end either at the middle caesura or at the end of the verse, occupying one half-line. Parry called this “unperiodic enjambment,” the situation where verse end coincides with a sense break though not with the end of a sentence. [31] To the extent, however, that sense breaks, cognitively defined, are more pertinent to the flow of the passage than any punctuation on our part, we might want to withdraw the term “enjambment” altogether for such situations. [32] Sometimes, however, matters are more complicated, as in the case of units 26 and 27:

These are two pairings of a subject (“ Hektor,” “the others”) and a name (“Skamandrios,” “Astyanax”). Each member of the pair contrasts with the corresponding member of the other pair; [33] the contrast is highlighted by the name being uttered after a metrical boundary. In 26 the boundary is the main caesura (indicated as “|”), and so verse-internal; in 27, on the other hand, the boundary is the verse boundary itself (indicated as “||”). The unit, in other words, is enjambing, not because any sentence is longer than the verse, but because the rhythmical properties of verse end are exploited to highlight an antithesis: a given word receives prominence by being uttered after a metrical boundary. [34]

In other cases the prime feature of the enjambing unit is not the prominence given to a given word but the rhythmical effect of the unit as such. Surveying longer stretches of Homeric poetry, we note that the cases of what Parry called “necessary enjambment” are quite frequent. [35] In a syntactic conception of enjambment, such cases, in presupposing a too elaborate sentential structure, are in principle troublesome for the orality hypothesis, for the reasons outlined above. On closer inspection, however, we note that these cases are not a matter of sentences becoming too long or syntax becoming too complex: the enjambment is a matter of a unit straddling the verse-boundary, running from the bucolic dieresis into the next verse. Furthermore, we observe something that is obscured by the statistics and percentage points that figure so prominently in the study of Homeric enjambment: this kind of “necessary” or “violent” enjambment tends to occur in clusters, creating areas of metrical turbulence at emotional high points in the narrative. An instructive example occurs shortly after Andromakhe’s first appearance, when she begins her speech: [36]

One of the most salient features of this passage are three strong sense-breaks at the bucolic dieresis, in units 3, 5, and 10. The first time this happens, in 3, an utterance starts that might be seen as complete in itself (οὐδ᾿ ἐλεαίρεις ‘don’t you take pity ’); it might seem that the sequel (παῖδά τε νηπίαχον ‘ little son’), is an adding unit in what Parry called unperiodic enjambement. Yet more is involved, for in the next line the same metrical position serves as sense break (unit 5), and this time the material at the other side of the verse-boundary belongs to the same speech unit as the material before. The same thing happens somewhat later, in unit 10, one of the relatively rare cases of a noun and its adjective being separated by the end of the verse. Thus a climax seems to build in this passage by means of a rhythmical profile that runs increasingly against the basic rhythm of the hexameter. Within the overarching metrical framework of the hexameter, a second rhythm is set up, striving temporarily against the basic rhythm yet at the same time reinforcing it. The final result is a highly effective fugal effect that regularly recurs at some of the more emotional moments of the Homeric performance. [37]

* * *

We will never know how Andromakhe’s words sounded before they were committed to writing; we don’t even know whether or to what extent they ever existed before writing. Nor does it matter. The effect that I have tried to capture and describe does not depend in any way on “writing” in any of the senses in which I have used the term in this chapter. It depends on the poet’s voice, whether we see it as a reader’s voice, a writer’s voice, or the voice of the textless singer of tales. Passages such as the one just presented have to be performed in order to reach their full potential. Their sophistication and enjambment is not syntactic as a subtle or complex interaction between the sentence and the verse; it operates in an entirely different dimension. This is the dimension of rhythm as the domain in which speech can be enhanced to the point at which it becomes poetry. The rhythmical finesse that we find in Homeric poetry does not make that poetry less “oral” in the sense that one of the tests for orality has not been passed; nor does it presuppose writing in our sense of written composition. What is important is that we have a text whose essence lies in being performed, and which gives us ample information on its oral conception.