publisher colophon


Rhythm and Rhetoric

All of poetry I consider to be and refer to as discourse with meter.

—Gorgias of Leontini, In Defense of Helen

It is strange how long it has taken the European literatures to learn that style is not an absolute, a something that is to be imposed on the language from Greek or Latin models, but merely the language itself, running in its natural grooves.

—Edward Sapir, Language

In the preceding chapters we have been concerned with what it means for language to be spoken, a question that presupposes the one which will occupy us in the remaining chapters of this book: what does it mean for speech to be special? In order to discuss the special, poetic features of Homeric discourse, we must first discuss the features of Homeric discourse as speech, in particular its segmentation into basic speech units that are cognitively determined.1 Accordingly, in the discussions to be presented in this third part of the study, we shall start each time from the cognitively determined intonation unit as it is observable in Homer. In the present chapter the subject approached in this way is meter; formulas in their semantic and metrical quality will be discussed in Chapters 7 and 8. The task will be to determine each time how one of the most basic properties of language in the spoken medium is enhanced, or stylized, into a poetic feature belonging to special speech, the discourse of the special occasion.

Distinguishing the two questions, the one pertaining to speech and the other to special speech, is useful as a method preventing us from approaching such matters as style and meter too early, and so assigning to them qualifications that apply only with hindsight, within a perspective of what is for us poetry. The boundary line between speech features and special speech features, however, cannot be drawn sharply: poetic features of Homeric style can be reduced to speech features precisely because speech features can easily become poetic.2 In the previous chapter we saw for example how framing, a common relation between intonation units, can result in ring composition and other “poetic” phenomena; in the present chapter we are concerned with how meter can result from the rhythmical regularization and streamlining of speech units.

The discussion of Homeric metrics along these lines does not take issue with the existing accounts of the dactylic hexameter or question their findings. Rather, it differs from them in the perspective chosen, in the direction from which some unquestionable metrical facts are approached. We are accustomed, whether consciously or not, to regard meter as a structure imposed on a discourse, a metrical form used for conveying the poetic message. We tend to view discourse units in a poetic text in terms of the metrical structure of the poem (verses, cola, and so forth). In the discussions that follow we shall try to reverse this perspective, not viewing discourse in terms of meter, but meter in terms of discourse. I propose to see meter not as a poetic form in itself but as emergent from spoken discourse with its typical spoken articulation. This argument will give us an opportunity to rethink some of the implications of the contrast between poetry and prose, and to search in everyday speech for the stylistic foundations of Greek literary texts.

Meter, Number, and Periodic Style

To start our discussion of rhythm and meter in Homeric speech, I return to the passage from the third book of Aristotle’s Rhetoric with which I began Part 2: the discussion of unperiodic and periodic style. Whereas our subject earlier was unperiodic style and the later versions of that concept (adding or paratactic style), we are now primarily concerned with periodic style, and the properties that Aristotle assigns to this mode of discourse (emphasis added):

λέγω δὲ περίοδον λέξιν ἔχουσαν ἀρχὴν καὶ τελευτὴν αὐτὴν καθ’ αὑτὴν καὶ μέγεθος εὐσύνοπτον. ἡδεῖα δ’ ἡ τοίαυτη καί εὐμαθής, ἡδεῖα μέν διὰ τὸ ἐναν-τίως ἔχειν τῷ ἀπεράντῳ, καὶ ὅτι αἰεί τι οἴεται ἔχειν ὁ ἀκροατὴς τῷ άεὶ πεπε-ράνθαι τι αὑτῷ· τὸ δὲ μηδὲν προνοεῖν εἶναι μηδὲ ἀνύειν ἀηδές. εὐμαθὴς δέ, ὅτι εὐμνημόνευτος. τοῦτο δέ, ὅτι ἀριθμὸν ἔχει ἡ ἐν περιόδοις λέξις, ὃ πάντων εὐμνημονευτότατον. διὸ καὶ τὰ μέτρα πάντες μνημονεύουσι μᾶλλον τῶν χύδην· ἀριθμὸν γὰρ ἔχει ᾧ μετρεῖται. (Rh. 1409a3 5 – 1409b8, ed. Kassel)

I call “period” an utterance with an inherent beginning and end as well as a length that can be beheld at a single glance. Such a type of discourse is not only pleasant, but also easy to learn. It is pleasant by the fact that it is the opposite of what is unbounded, and because the listener at every moment has the idea of having hold of something, by the fact that every moment is bounded in itself. For having no anticipation of what is to come or not completing anything is unpleasant. It is easy to learn because it is easy to recall. This is because periodic discourse has number, which of all things is easiest to recall. This is why all people memorize metrical discourse more easily than language that is poured forth. For it has number by which it can be measured.

We usually discuss periodic style in terms of the balanced syntax and stylistic subtlety that come with hypotactic construction, defined in opposition to parataxis. Aristotle, however, appears to have a quite different conception of periods.3 Using the term “period” to designate what are for us parts of periods, he assigns to these smaller units properties that do not seem to be directly applicable to our syntactic conception of period. The periodic style is “pleasant,” according to Aristotle, because of its boundedness, which is a matter of hearer’s anticipation: periodic style, as opposed to unperiodic style, gives the hearer a sense of what will come next. From Aristotle’s wording it is not immediately clear whether this anticipation is a matter of the relation between periods or of the period internally. In any case, the second property mentioned, that something in periodic style is “easy to learn,” applies to the latter possibility. Aristotle says of periodic style that “it is easy to learn because it is easy to recall” (eumathḕs dé, hóti eumnēmóneutos), the reason being that periodic discourse has number, which is “of all things easiest to recall” (pàntōn eumnēmoneutótaton). For Aristotle, number is another kind of pleasant limitation,4 and hence the pleasures of the periodic style, which derive from the styles boundedness, are related to the ease with which periodic speech is remembered, for that ease comes from the bounded nature of its rhythm or number.

Metered poetry also belongs to this category of number, as Aristotle mentions in passing. This move may be unexpected for us in a discussion of what is for us prose, but it becomes less surprising when we realize that rhetorical prose in antiquity is less a mode of written communication than a specific speech genre that is continuously defined with respect to poetry: while the poetic speech genres have number and rhythm to such a degree of fixity that one can speak of meter, the periodic style of rhetorical prose is periodic in its being less strict and fixed than poetic meter, but rhythmical and numbered all the same. For Aristotle, then, periodic style seems to be less a matter of syntactic composition and production than of the experience of rhythm—and therefore a matter of delivery and performance. Aristotle actually uses terminology that for us, paradoxically, seems more appropriate for the constraints under which oral composition and/or recall in performance has to take place.5 But if periodic discourse is discussed in terms suggesting rhythmical anticipation and memorization, what are we to do with the allegedly unperiodic style of Homeric metered poetry, where rhythm and number are, if anything, even stronger?

Is Homeric style periodic? The question seems a contradiction in terms when we define period in the modern way. As we saw in the previous chapters, Homeric discourse is by no means simply paratactic, but that is not to say that Homeric syntax is hypotactic. The question becomes more interesting, however, when we adopt Aristotle’s different perspective and elaborate on it, viewing rhetorical periodic discourse not in terms of syntax or style but in terms of speech, as a special discourse for a special occasion. The opposition between Homeric poetry and classical rhetoric then becomes not so much a matter of style, of the Kunstsprache of Homeric epic as opposed to the Kunstprosa of the classical period, as of different performance genres, each departing in its own way from ordinary everyday discourse.

In this chapter I explore some of the possibilities of viewing Homeric discourse within this rhetorical framework. Taking the categories poetry and prose somewhat less for granted than they are usually taken, I will argue that both Homeric poetry and classical rhetorical prose are, each in their own specific and very different ways, the rhetorical enhancement and manipulation of the basic properties of ordinary speech.6 Both are special speech, based on strategies that are reserved for special performance occasions and meant, in a truly rhetorical sense, to have a special effect on an audience. The main thrust of the argument is that the stylistic opposition between periodic and unperiodic may not be the most meaningful way to bring out the differences between Homeric poetry and classical rhetorical prose. Rather, we might say that both are periodic in their own very different ways.

The periodic nature of Homeric discourse, I will argue, lies in the interaction between the speech units and meter. In this kind of discourse, in which periods are defined in terms of meter, the number quality of the discourse units is so strong that it determines not only the units as such, but also the way in which they are related to each other as rhythmical units. This genre of special speech will be opposed to the prose type of discourse, in which number is merely a matter of the rhythm of speech units taken by themselves: an important property, but still subordinate to the way in which the speech units follow each other in the flow of rhetorical discourse. Let us now consider how both types of periodicity can be best described and how they achieve their specific rhetorical effects.

Rhythm in Speech

We saw that for Aristotle number, or rhythm, is an important factor in the learning or recall of a discourse. But rhythm, as a constraint that facilitates memorization, would not be effective if it did not operate in concert with the general possibilities and limitations of human consciousness that we discussed in the previous chapters. On this basis we can view the intonation unit of ordinary spoken discourse as the proper locus for rhythm, and approach Homeric meter from the vantage point of speech. In doing so we are encouraged by the general observation that rhythm is not an external factor, superimposed on language from outside; language has its own rhythm, which can be strengthened, regularized, and standardized to the point at which we can speak of meter. Rather than separating poetry from language as art from life, then, the metrical factor integrates poetry within speech, in ways that turn the discussion of meter in the present chapter into a complement of the discussion of consciousness and cognition in the previous ones.

No speaker is entirely at the mercy of the cognitive limitations that I described in Chapter 3, and cognition is not the only factor that makes speech what it is, at least speech as the object of stylization in the form of special speech. It is true that spoken discourse may be best described as a process, revealing some of the properties of consciousness as a process or flow; but that does not mean that this process is involuntary, having only the limitations of the processing consciousness as its constitutive features. In fact, speech is purposeful behavior no less than it is constrained by cognition, frequently involving deliberate choices as to presentation and effect. To bring out this important aspect of speech, I will speak of rhetoric in this chapter, as the necessary counterpart of the cognition that I examined in Part 2. Cognition and rhetoric, the latter concept to be understood in a broad, pretheoretical sense, can be seen as opposite but interrelated forces.7 It is the interplay of cognitive and rhetorical features, in varying ratios, that defines the style of most ordinary spoken discourse. And a discourse is rhetorically more sophisticated to the degree that the intonation units of ordinary speech are consciously and deliberately manipulated, in various ways, but without their losing their cognitive role in the production and reception of the discourse in question.

Even the most casual, unpremeditated discourse displays rhetorical features that are revealing for the strategies that we will see later on in more rehearsed and sophisticated discourses. Consider the following example from Chafe’s corpus of data on spoken discourse, the climax of a conversational narrative:8

  1. … And there were these two women,
  2. .. hiking up ahead of us.
  3. … [r.s] And you sort of got,
  4. to a rise,
  5. and then the lake,
  6. was kind of right there,
  7. where we were gonna … camp.
  8. … And the two of them,
  9. .. got to the rise,
  10. .. and the next minute,
  11. … [0.9] they just .. fell over.
  12. .. Totally.

This highly informal fragment, consisting of twelve intonation units, is a curious mix of cognitive and rhetorical features. The rhetorical features, which concern us here, occur at the end of the passage, in the last three units.9 There are two long pauses in the fragment, one before unit c and the other between units j and k. The two breaks are similar in length, but of an altogether different nature. Unit c, along with units d-g, is framing and orienting: it sets the scene for the event to be narrated. The pause preceding unit c seems to reflect the mental effort connected with the activation, the visualization of the scene; it is followed by four swiftly delivered units, as if the substance of those units had been “booted” during the pause. For this reason we may call the pause cognitive. The pause after unit j, on the other hand, can be called rhetorical: instead of merely reflecting mental effort, it is used as a device to create suspense at the climax of this little narrative. In fact, the pause is part of a presentation strategy which also involves the unit preceding the pause. This unit was uttered at a higher pitch, with more volume, and above all, at a slower pace than the previous units (almost in a chanting manner), a phonetic realization of which the pause is an integral part and the main reason why it can be called rhetorical.10

Part of the rhetorical articulation at the end of the fragment is also unit l, which has something in common with the type of intonation unit that verbalizes a detail added to the previous unit.11 Yet in the present context it seems to have a more complex function. Apart from being the verbalization of a detail added to the previous moments verbalization (a detail which stresses the extraordinary nature of the event), this unit seems to have a rhetorical, or more precisely, a rhythmical function. Being uttered just after the peak of the narrative, it enhances the salience of the peak event (unit k), by modifying it across an intonational boundary (indeed a sentence boundary, marked by the full stop that signals falling intonation). In the two previous units, the discourse had developed into a process with its own pace and articulation; unit j prepares and leads up to unit k, and the l-unit, in counterbalancing the preparatory unit j, provides a rhetorical addition to the peak unit. Its presence seems to be best explained by the fact that its absence would have been undesirable from the point of view of the rhythmical relations between units.12

Even in relatively simple and informal cases, then, the transformation of consciousness into speech may result in an object with properties and processes of its own, besides those deriving from the flow of consciousness. For the purposes of the present argument it is opportune to focus on those rhetorical properties that can be discussed under the general heading rhythm. Rhythm is a property of any spoken discourse. Yet its importance and deliberate use depend on the degree to which a narrative is rehearsed and rhetorically presented.13 No discourse is entirely devoid of rhythm, but in some discourses and discourse types, rhythm is more important than in others, testifying to the varying ratio of cognitive and rhetorical features. In the remainder of this chapter I am concerned with the way in which rhythm as a feature of special speech may come to stylize the cognitively determined features of ordinary speech.

Rhythm and the Remote Consciousness

When we move from informal narratives like the one just presented into more rhetorical territory, an instructive example of the interaction of cognitive and rhetorical (rhythmical) features is the discourse and performance of American folk preachers, described by Bruce Rosenberg as a genuine oral tradition,14 and known to a larger audience through the oratory of Martin Luther King, Jr., and Jesse Jackson. A folk sermon is typically chanted, at least in its more intense and emotional stages, and presented in short, rhythmical units. The chanting is improvised, but due to the recurrence of each sermon in a preacher’s practice as well as to the formal and official stance of the speaker, there is obviously more rehearsal, planned organization, and professional experience than in informal dinner-table conversations. The following passage is cited by Rosenberg as an accurate description of this type of discourse and its performance:15

Rev. Ratliff begins his sermon in normal, though stately and carefully measured prose. As he gets into his subject, he gradually raises the intensity of his delivery (though with well-timed ups and downs). About one third of the way into his sermon the prose has verged into a very rhythmical delivery, punctuated into periods (more or less regular) by a sharp utterance which I suppose might be called a vehement grunt. I haven’t timed these periods, but I would guess that they fall about every three seconds, sometimes less. Within the rhythmical framework, the rises and falls eventually build to a climax when he lapses into a sort of chant, still with the same punctuation, but with a recognizable tonic [tonal center]. Some of the congregation (who respond ad libitum throughout) here lapse into humming along with him.

This description evidently applies to phenomena that are closely related to Chafe’s two-to-three-second intonation units. The “periods” described, however, seem to be more pronounced and recognizable than the intonation units of ordinary speech, both intonationally and rhythmically. More-over, the rhythmically enhanced units apparently play a more important role than just the accommodation of the flow of discourse to the flow of consciousness in the minds of the listeners: delivered at the more emotional stages of a sermon, or when the psychological conditions are right, they actually invite the audience’s active participation in the flow of discourse. The following transcription of such a performance will serve as a fairly typical example:16

  1. John said
  2. I… I
  3. I saw four beasts
  4. One with a face
  5. Looked like a calf
  6. Representin’ patience
  7. And endurance
  8. ’Nother beast I saw
  9. Had a head like a lion
  10. Representin’ boldness
  11. And confidence
  12. ’Nother beast I saw
  13. A face like a man
  14. Representin’ wisdom
  15. An’ he had knowledge
  16. ’Nother beast I saw
  17. Looked like a bald eagle
  18. Ain’t God all right?

This catalogic passage, an instantiation of the four-beast theme from the Apocalypse, exemplifies the interplay of cognitive and rhetorical features when a folk preacher successfully draws on the resources of his tradition. Each unit represents a separate idea on which the performer focuses, and their sequence represents the flow of these ideas through time, as well as their interrelationships, both in the mind of the performer and in the minds of the audience. There is appositional or adding syntax in the representin’-units f, j, and n, but also in the lines preceding, which verbalize an added piece of detail as an independent clause.17 But sharp demarcations between the units and distinct phonetic contour are not as such reflexes of cognitive constraints: a positive, rhetorical factor is at work, operating upon the basic units of speech in their preferred length and in their typical syntactic inter-relationships. This factor consists of the performer’s presentation strategy to turn intonation as the main physical property of his speech units into rhythm,18 a strategy that not only affects the internal constituency of the units and their relative timing, but also their length.

But rhythm is not an inherent, automatic property of this sermon, or of the speech genre as a whole. The delivery of rhythmical speech units is a matter of performance, and no two performances are identical. Rosenberg actually cites a different, much less rhythmical version of the same theme (with more cognitive features and hesitation phenomena, we may add), which was delivered under much less favorable performance conditions.19 Apparently the amount of rhythm—in our terms, the ratio of cognitive and rhetorical features—varies from performance to performance, according to audience response and the performer’s inspiration at the moment. Without a fixed text serving as norm for future performances, the rhythm of the African-American sermon remains a matter of performer-audience interaction, and hence dependent on the chemistry of the moment.

It is important to emphasize at this point that rhythm in the sermon preachers’ discourse is not a rhetorical manipulation of any preexisting ordinary discourse. As we saw, no discourse is entirely devoid of rhythm or rhetoric, and the African-American sermon differs not in substance but in degree from more casual discourses. The tradition on which the preachers draw, with its roots in biblical rhetoric and African-American Baptist culture, departs from ordinary speech in the rhythmical regularity of speech units at a sermon’s more intense moments, but it is not for that reason wholly different from speech. The preachers’ chanted lines have a poetic quality precisely because they are delivered in the interactive context of a performance. In such a context, the boundaries between cognition and rhetoric, speech and poetry, may become irrelevant to the point at which we may speak of poetry in speech. It takes text, and a textual conception, to separate what is indissolubly connected in a performance, and to isolate poetry as something removed from speech.20

Rhythm as a feature of a performance is in a number of ways a complicating factor in an analysis that considers special speech, just as any speech, to be the actualization of the speaker’s consciousness. Rhythmical discourse does not merely reflect the speaker’s flow of consciousness, but at the same time in a way directs the flow, which has properties and dynamics of its own that are highly conducive to memorization and recall. Rhythm, in fact, increases the impact of speech as an event. The speech may derive from the speaker’s cognitive processes, but it is at the same time an independent process in which the speaker himself can participate, given the appropriate conditions, both private or psychological and public, relating to the dynamics of the performance.21

It is this potentially ambiguous relation between rhythm, consciousness, and speech that accounts in part for the fact that many performers in traditions of special speech around the world experience their discourse as deriving from a consciousness other than their own, their role in the performance as being that of an interpreter or mediator, and their behavior as being divinely inspired. Rhythm, in other words, contributes to what might be called a dislocation of consciousness: the speech produced is not the present speaker’s responsibility but something with which a remote authority is credited, an authority located beyond everyday experience and the source of immutable knowledge and truth.22

Rosenberg’s preachers consistently claim that their power to produce rhythmical chant derives from God,23 and in the Homeric context, it is impossible not to think of the Muses. While other epic traditions stage their performers as telling what they heard, in the Greek context the singer plays the role of an eyewitness.24 This is possible because the performer is a theîos aoidós ‘godlike singer’ and hence the favorite of the Muses, the ultimate eyewitnesses, who were present on the battlefield of Troy where the epic events were enacted.25 It is their remote, divine consciousness that the epic poet makes present in the context of the performance. Consider in this connection the words of the Ithakan bard Phemios in the Odyssey:

αὑτοδίδακτος δ’ εἱμί, θεὸς δέ μοι ἐν φρεσὶν οἴμας παντοίας ἐνέφυσεν ·

(Od. 22.347–48)

I am self-taught and for me in my mind a god made song-paths of all kinds grow.

These words are often compared with the often cited claim of a Kirghiz bard recorded by the Russian folklorist Vasilii Radlov: “I can sing any song whatever; for God has planted the gift of song in my heart. He gives me the word on my tongue, without my having to seek it. I have learnt none of my songs. All springs from my inner self.”26 The apparent opposition between self and god in these passages has been taken as reflecting a distinction between form (the formulas that are the poet’s own contribution to the poem) and content (the god’s contribution).27 But the terms “form” and “content” are alien to the self-presentation of these singers.28 And with the collapse of the form-content distinction, the notion that the poet’s self and the divine are opposing factors in the poetic process loses much of its attraction.29 Rather, what is at stake, among other things, is a moment in the process of verbalization, the transformation of the stream of private consciousness into a stream of public and rhythmical speech. The usual account of speech as deriving from consciousness is insufficient here, for the singer’s consciousness not only produces the speech but is also propelled forward by the rhythmical movement of the language. What springs from the self is thus both larger and stronger than the self. The power of such a speech can then be said to originate from a source that is neither opposed to the speaker’s consciousness, nor identical to it. And this explains the speaker’s claim that a song which is planted by a god springs from the singer’s own self.

We could return here to the discussion of rhythm as a rhetorical presentation strategy and move from Phemios and his self-presentation as bard to the practice of Homeric rhythm. To do more justice, however, to the complexity of the latter, and above all its rhythmical periodicity, we have to turn briefly to the rhythm of ancient rhetorical prose. According to Axistotle, as we have seen, the periodicity of this genre of discourse consists in its having number and in its allowing the listener to “anticipate” (pronoeîn). At the end of the chapter, we will be comparing these two types of period to each other.

From Cicero to Homer

Rhythm, in a more prosaic vein than contact with the divine, belongs to the many involvement strategies at the disposal of speakers, both casual and formal, that are discussed by sociolinguists.30 But the discourse analyst of the late twentieth century is by no means the only one to discuss these matters. Students of classical rhetoric will recognize in the sociolinguists’ involvement strategies the figures of style from ancient rhetorical theory, the stylistic embellishment of Greek and Latin rhetoric and poetry. But while these figures and tropes have been treated as art, and the speeches in which they occur as Kunstprosa that is removed from the naturalness and artlessness of ordinary speech,31 the very fact that ordinary speech abounds in figures suggests that the art of Greek and Latin rhetoric is effective precisely because it draws on the common strategies of everyday speech.32

In fact, the ancient theorists of the art of public speaking themselves had a conception of rhetorical style that, though in many ways a forerunner of the modern conception of Kunstprosa, was still quite different from it. Being much closer to what may be called discourse analysis than some of their modern students, these writers consider the art of formal public speaking as drawing on, not separated from, the everyday and the ordinary. Style is for them a matter, not only of aesthetics—the static form of a discourse as text—but also of pragmatics: their central concern is a more dynamic conception of discourse as behavior, based on deliberate presentation strategies and a skillful manipulation of the properties of ordinary speech that are essential for understanding and rhetorical success.

One of the most important of these properties is the segmentation of speech into short units, whose cognitive necessity and rhetorical potential we have already seen. Basic speech units are the core of what is “nature” in rhetorical discourse and at the same time are the foundation and starting point of what is “art” in that speech genre. Consider what the Roman orator and rhetorical theorist Cicero has to say on this issue, in a discussion of style that is remarkable throughout in its insistence on sound, rhythm, and other performance-related matters:

clausulas enim atque interpuncta verborum animae interclusio atque angustiae spiritus attulerunt: id inventum ita est suave, ut, si cui sit infinitus spiritus datus, tamen eum perpetuare verba nolimus; id enim auribus nostris gratum est, quod hominum lateribus non tolerabile solum, sed etiam facile esse posset. Longissima est igitur complexio verborum, quae volvi uno spiritupotest. (De orat. 3.181)

It was failure or shortness of breath that originated periodic structure and pauses between words; but once invented, this [segmentation] proved so attractive that even if there were a person endowed with unlimited powers of breath, we would still not want this person to deliver an uninterrupted flow of words. For our ears are adapted to what is not merely endurable but also easy for the lungs. The longest stretch of words, therefore, is that which can be completed in one single breath.

For Cicero there are pulmonary constraints on the flow of discourse, a physical necessity resulting in observable and expected breaks in the flow of speech in the performance and yielding a segmentation into relatively short units.33 Today we would not argue that breathing is solely responsible for breaks in a discourse and would more likely consider it as synchronized with the segmentation resulting from cognitive constraints.34 For Cicero and from his point of view, however, there is another happy synchrony. Being one of the things in nature that bring beauty and dignity by their very usefulness and necessity,35 the primary units of discourse are the source of rhythm, which more than anything else characterizes rhetorically enhanced discourse as the skillful manipulation of what is natural to speech. It is the rhythmical aspect of speech units, both in their length and internal constituency and in the movement from one unit to the other, that turns constraint into a positive source of involvement by capturing and directing the attention of an audience. Again, it is instructive to cite Cicero on this point:

et, si numerosum est in omnibus sonis atque vocibus, quod habet quasdam impressiones et quod metiri possumus intervallis aequalibus, recte genus hoc numerorum, dummodo ne continuum sit, in orationis laude ponetur. Nam si rudis et impolita putanda est ilia sine intervallis loquacitas perennis et profluens, quid est aliud causae cur repudietur, nisi quod hominum auribus vocem natura modulatur ipsa? quod fieri, nisi inest numerus in voce, non potest. Numerus autem in continuatione nullus est; distinctio et aequalium et saepe variorum intervallorum percussio numerum conficit, quern in cadentibus guttis, quod intervallis distinguuntur, notare possumus, in amni praecipitante non possumus. (De orat. 3.185–86)

But if this element of rhythm is in all sounds and voices, characterized by certain beats and measurable by its regular intervals, then its presence in discourse, provided it does not occur without interruption, will be a thing worthy of praise. For if a continuous flow of words has to be considered rough and unpolished, is there a better reason to reject it than the fact that nature herself modulates the voice for the ears of humankind?—a thing that would be impossible unless the voice inherently contains an element of rhythm. But in an uninterrupted flow there is no rhythm. It is segmentation and a beat characterized by equal but often varied intervals that creates rhythm. Rhythm is what we notice in falling drops of water, because they are separated by intervals, not in a fast flowing river.

Cicero goes on to point out that the rhythmical movement of the membra ‘limbs’, the short units of discourse, needs considerable management if it is to be felicitous in all regards. Apart from the fact that the rhythmical manipulation of units is an important factor in the creation of certain moods in discourse and in the adaptation of the flow of discourse to the various oratorical genres and styles, there are two major considerations in the rhythmical articulation of discourse. First, the rhythmical movement of rhetorical discourse should not be too close to poetic meter in the internal rhythmical structure of the units, nor in the rhythmical relationships between the units. This stylistic constraint on rhetorical discourse remained valid throughout the history of ancient literary and rhetorical criticism.36 Second, the rhythmical articulation of rhetorical discourse has to support the syntactic articulation in order to attain the desired periodic structure, in which not only rhythm is brought to completion, but also syntax and thought. The units near the end of a period, for example, have to be longer than the preceding ones in order to avoid truncation and to secure a pleasing ending.37

The two considerations work together in establishing rhetoric, not as an artful or even artificial variant of what is for us prose, but as a particular genre of special speech. In its specific deviation from ordinary speech, this genre distinguishes itself most clearly from other, more traditional speech and performance genres, such as the Homeric one. In carrying out the natural rhetorical strategy of adding rhythmical articulation to intonation as a physical property of speech units, the ancient rhetor is faced with the existence of the easily identifiable rhythms of the established poetic genres. Avoidance of these may have been a matter of taste and stylistics from the fourth century b.c.e., when rhetoric had established itself as a discipline with intellectual dynamics of its own.38 But the earliest stages of the development of ancient rhetoric have to be situated in the fifth century b.c.e., a time less concerned with rhetorical or poetic theory than with actual performance. And in this climate, awareness of the rhythmical possibilities of one’s speech may have been more a matter of the awareness of the existence of rival performance genres than the intertextual differentiation of one literary style from the other.

The fifth-century rhetors and sophists Thrasymachus and Gorgias come to mind in this connection. The former is traditionally credited with the invention of rhythmical Kunstprosa,39 and the latter with the introduction of figures of speech related to the length and configuration of speech units.40 These two orators are the first in a tradition that defined, analyzed, and designed rhetorical discourse with an ear to poetry. But instead of introducing poetic, metrical elements into prose, as has often been assumed from antiquity onwards,41 they seem to have been more concerned with the development of a performance genre sufficiently close to poetry (i.e., special speech in performance) to be rhetorically effective (i.e., have a similar emotional or psychological effect), yet sufficiently different from the way in which poetry deviates from ordinary speech to rank as a separate genre.42 Consider, for example, the following excerpt from Gorgias’s epideictic speech In Defense of Helen, a demonstration of the power of logos that is based on the mythical “case” of Helen, whose abduction by Paris caused the Trojan war:

ἐγὼ δὲ βούλομαι λογισμόν τινα τῷ λόγῳ δούς τὴν μὲν κακῶς ἀκούουσαν παῦσαι τῆς αἰτίας, τοὺς δὲ μεμφομένους ψευδομένους ἐπιδείξας καὶ δείξας τἀληθὲς παῦσαι τῆς ἀμαθίας. (Hel. 2)

What I want is to provide an argumentation in my speech so as to keep her who is held in bad esteem from accusation, and to demonstrate the lies of those who blame her, and furthermore to give a demonstration of the truth and to keep [them] from ignorance.

Presented in the usual way, this passage is an example of early Greek prose, characterized by an antithetic contrast expressed by the particle pair méndé. But this method of written presentation obscures the fact that the passage in performance must have contained silence and hence exhibited the fragmented quality we discussed in the previous chapters. It consists of short units that match the ideal cognitively determined length of the intonation units of ordinary speech, but which in their rhythmic profile resemble the rhythms of poetry. Following Norden,43 I now present the passage as a sequence of intonationally and rhythmically marked lines rather than as continuous prose:

a. ἐγὼ δέ βούλομαι

But as for me, I want,

b. λογισμόν τινα τῷ λόγῳ δούς

providing argumentation in my speech,

c. τὴν μὲν κακῶς ἀκούουσαν

her who is held in bad esteem

d. παῦσαι τῆς αἰτίας,

to keep [her] from accusation,

e. τούς δὲ μεμφομένους

and as for them who blame her,

f. ψευδομένους ἐπιδείξας

having demonstrated their lies,

g. καὶ δείξας τἀληθὲς

and given a demonstration of the truth

h. παῦσαι τῆς ἀμαθίας.

to keep [them] from ignorance.

For ancient critics, poetry is a type of discourse conforming to a regular metrical sequence, and by that definition this passage from Gorgias is prose. But because, like poetry, Gorgias s In Defense of Helen presents rhythmically enhanced speech units in performance, it is certainly not prose. Some of the units of the passage, in fact, are almost identical to the recognizable rhythmical units of the various poetic genres. Units a and c, for example, strongly resemble iambic rhythm (ᴗ - ᴗ - ᴗ - and - - ᴗ - ᴗ - - -), and units e-f are dactylic in feeling (- ᴗ - ᴗᴗ - and - ᴗᴗ - ᴗᴗ - -) and almost hexametrical in their relation to each other. The passage, in short, is speech rhythmically enhanced in such a way as to resemble poetry. It has been partly designed as an imitation of the way in which poetry, as special speech, distinguishes itself from ordinary speech.

Besides rhythmic articulation, we must also consider the syntactic arrangement of the various units. We saw in the previous chapter that the progression of speech units in Homeric discourse is often a matter of framing: units frequently provide context for speech to come. In and by themselves, however, Homeric speech units tend to be syntactically and semantically autonomous: they agree with each other grammatically, with-out there being government of one by another.44 In Gorgias’s discourse, on the other hand, and in rhetorical prose generally, the framing is a matter of syntactic government of one unit, or set of units, by another. Instead of the continuation or addition I discussed in Chapters 4 and 5, the most characteristic relation between two units is complementation; units tend to be syntactically incomplete or completing, and the way in which one unit anticipates another is thus a matter of syntactic construction.

The particles mén and in units c and e, for example, rather than marking the role of independent clauses in the flow of discourse, as in Homer,45 signal an antithesis between two ideas within the framework of an overarching construction, in accordance with the use of mén that is usually singled out as central in Greek grammar. This syntactic intricacy, in which looking forward on the path of speech is “syntacticized” to a considerably higher degree than in Homer (note the parallelism between units d and h), is the hallmark of the periodic style of classical rhetoric, of which Gorgias’s passage is an early example. In its mature form, the art of rhetorically manipulating the attention flow of the listener reached the point where the units (Latin membra, Greek kola) of discourse are either protases or apodoses in the widest possible sense, in that they either create syntactic expectations or give “what is due” in fulfilling them. It is to this syntactic periodicity and anticipation that the rhythmical articulation of speech units in their rhetorical form is ultimately subservient, in both the theory and the practice of rhetorical discourse from the fourth century b.c.e. onwards.

To consider Homeric discourse unperiodic, in contrast to the periodic syntax of a Gorgias, is problematic and unsatisfactory, as we have seen in the previous chapters. Homeric discourse looks ahead in its progression no less than does rhetorical discourse. Still, we may not want to consider this looking ahead as anticipation in the periodic sense: in spite of all the framing and orientation, there is no period to be completed, and no over-arching syntactic construction with respect to which the ordering of speech units is arranged. Instead, there is an entirely different way in which speech units in Homer are anticipating each other in the periodic sense: the dimension of meter. Units in Homer are defined not with respect to the completion of a syntactic whole, but with respect to the completion of a metrical whole. In other words, if we take the rhetorical periodic sentence as one highly complex end point in the marking and manipulation of speech segments, then it appears that Homeric discourse with its metrical period, is another equally complex end point, the perfection of an entirely different genre of special speech.

From Rhythm to Meter

Retracing our steps away from periodicity and anticipation for a moment, let us go back to a type of discourse like that of the American folk preachers discussed above, where, under the right performance conditions, a ratio of cognitive and rhetorical features may obtain that differs from ordinary discourse: the speech rhythm becomes more regular, to the point at which we may speak of meter. Typologically, such a discourse could be seen as a point on a scale ranging from maximum cognition to maximum rhetoric; this scale can also be considered diachronically, as a gradual development from unmarked, ordinary speech to marked, special speech;46 at a given point on this scale we might envisage the differentiation of two deviations from ordinary speech, the one leading to the syntactic periodicity just discussed, and the other to periodicity in the form of a metrical period to be discussed now.

Regularized rhythm is a rhetorical strategy to emphasize, as I have argued, the boundedness of intonation units, in order to accommodate the discourse to the listeners’ consciousness and to stimulate the participation of the audience in the flow of discourse. Yet at some point rhythm ceases to be simply a property of intonation units and becomes subservient to something else. This happens when the rhythmical movement of the units becomes so regular as to turn into a period, consisting of indefinitely repeated verses or rotations. This is the point at which the rhythm emergent in discourse becomes meter.47 Thus meter is not something superimposed on language, a form that exists independently of it; meter emerges from language as part of the process by which special speech emerges from speech. A useful formulation of this understanding of meter is provided by Nagy: “At first, the reasoning goes, traditional phraseology simply contains built-in rhythms. Later, the factor of tradition leads to the preference of phrases with some rhythms over phrases with other rhythms. Still later, the preferred rhythms have their own dynamics and become regulators of any incoming phraseology.”48

The simplest case of meter, and minimally different from mere rhythmically streamlined intonation units, is the coincidence of the metrical period with the basic segments of speech and performance. This type of meter, which of all the metrical systems is closest to speech and in which rhythmical properties belong just as much to the discourse units (formulas) as to the meter, is exemplified in oral traditions all over the world. In the context of the older Indo-European languages, Rig-Veda verse may serve as an example, and in later times we find a variety of medieval traditions and verse types. Other examples, some of which are modern, include the Finnish Kalevala, as well as the verse of the South Slavic tradition investigated in the 1930s by Parry and Lord.49 All these cases may be characterized, for the purposes of the present discussion, in terms of adding, both in the usual sense of adding style, applying to the absence of syntactic anticipation, and in the sense that each new speech unit added to the previous one is as such a new instance of the metrical period.

Homeric discourse and metrics is much more complex than this, and also an important step further removed from ordinary speech in its rhetorical sophistication. The Homeric metrical period, the dactylic hexameter, is much longer than speech units usually are, and this turns the movement from unit to unit into something much more complex, metrically and rhetorically, than the mere recurrence of an identical period. Instead of coinciding with the metrical period, speech units in Homeric discourse are in their rhythmical profile subservient to it, as part of an ongoing rhythmical flow, and this creates anticipation and periodicity, not in the syntactic sense of rhetorical period but in the equally rhetorical sense of metrical period.

It has often been pointed out that the dactylic hexameter is a long verse, with one, two, even three caesuras, which accordingly displays an internal structure that accommodates two, three—according to some even four—cola. Much discussion has been devoted to the exact status of these caesuras and cola: are they metrical or syntactic?50 It may not in the end make much difference whether we speak of metrical units or of syntactic units seen in terms of meter. The same rhythmical properties will be assigned to the verse as to its manifestation in language. As long as we consider meter the perspective and starting point, it may not even make much difference whether we see the hexameter as a unit of discourse in itself, or as a mode of discourse, a rhythmic principle regulating the flow of discourse. What is of interest here is the specific nature of this flow of discourse, treated as a progression of cognitively determined speech units. It is this cognitive flow, I submit, on which we must base our discussion of the rhetoric of the Homeric hexameter.

Instead of speaking of discourse in terms of meter, then, we are dealing with meter in terms of discourse, as part of our discussion of the emergence of poetry out of speech. In terms of cognition, the hexameter cannot be an original discourse unit: it is simply too long to be grasped in its entirety by the poet’s and listener’s consciousness. Instead, I propose, the hexameter is a matter of rhetoric, of the deliberate manipulation of speech units for the purposes of special speech in performance. The Greek epic tradition, at an early date, must have developed a verse (in the literal sense of “period”: something that returns to its beginning) that deliberately exceeded the span of human consciousness, creating a complex universe of discourse that is suitable for the reenactment of complex epic stories and the words of their characters.51 The exact metrical details of this process will probably remain forever in the dark, but the origin of the hexameter from smaller units is a very plausible scenario, if not an inevitable one from a cognitive point of view. Instead of a coalescence of two shorter verses, however, we would have to think of an increasingly elaborate rhythmical interdependence of the basic units of speech.52

The space opened up by the long metrical period allows the poets to move in various ways from one period to the other, thus producing the complex colometry that was taken up by subsequent hexameter poets. The least complex situation is exemplified by a passage that served in Chapter 3 as an example of Homeric adding style: the metrical period is realized as two intonation units that have now become metrical units, conforming to the two halves of the verse divided by the middle caesura, which occurs either immediately after the third long position (the so-called penthemimeral caesura) or, when the third foot is a dactyl, after the first short syllable of that foot (the feminine or trochaic caesura):53

a. || Πρῶτος δ’ Ἀντίλοχος

And first Antilokhos,

b. | Τρώων ἕλεν ἄνδρα κορυστὴν

of the Trojans he took a helmeted man,

c. || ἐσθλὸν ἐνὶ προμάχοισι,

valiant among the foremost fighters,

d. | Θαλυσιάδην Ἐχέπωλον·

Thalusias’s son Ekhepolos.

e. || τόν ῥ’ ἔβαλε πρῶτος

He first struck him,

f. | κόρυθος φάλον ἱπποδασείης,

on the crest of his horse-haired helmet,

g. || ἐν δὲ μετώπῳ πῆξε,

and he planted [it] in his forehead,

h. | πέρησε δ’ ἄρ’ ὀστέον εἴσω

and it pierced right through the bone,

i. || αἰχμὴ χαλκείη ·

the bronze spearpoint,

j. | τὸν δὲ σκότος ὄσσε κάλυψεν,

and darkness covered his eyes,

k. || ἤριπε δ’ ὡς ὅτε πύργος,

and he fell as when a tower [does],

l. | ἐνὶ κρατερῇ ὑσμίνῃ.

in the tough battle.

(Il. 4.457–62)

In this fragment, the movement from the verbalization of one focus of consciousness to another in the adding style is subjected to a metrical cohesion between pairs of units that shows Homeric discourse at its rhetorically least marked and least complex: each unit that begins the metrical period (units a, c, e, g, i, and k) is followed by a unit completing it (units b, d, f, h, j, and l), and each pair is separated by what in terms of hexameter metrics is the middle caesura. This pattern of rhythmical segmentation often yields one of two paired rhythmical units (minimal strophes or distichs), depending on where the middle caesura falls:



In the example above, units c-d, units g-h, and units k-l are of the second type, and the others are of the first type. In either case, the second unit of the line complements the first. So between the two units (cola), there is a mutual expectancy that is not unlike that between a protasis and an apodosis in a rhetorical period: the one fulfills what the other promises. In the next chapter we will see that in certain central cases this expectancy is no less a matter of the meaning of the two units.

The two-colon period just described is the default case of rhythmical articulation in Homeric discourse. It displays controlled variation not only in that each complementing unit has a different rhythmical profile from its opening unit, but also in the way in which entire periodic distichs (the two units of a verse taken together) relate to each other. When any two adjacent verses are considered together, the result we observe is a more or less tightly organized interplay of four rhythmical profiles. Whatever role the middle caesura has played in the origin of the hexameter, its rhetorical role is beyond question: in either of its two possible realizations, it is the primary resting point in the movement from the beginning to the end of the metrical period.

But Homeric discourse frequently moves beyond this level of rhythmical and rhetorical complexity. Other subdivisions of the time span of the metrical period are possible, yielding cognitive breaks after either the trithemimeres (i.e., a break in the middle of the second foot) or the hephthe-mimeres (i.e., a break in the middle of the fourth foot), or at both places:


In the last case, the characteristic result is what Kirk has called a rising threefolder, a division of the time of the metrical period into three units of increasing length:54

|| τὸν δ’ εὖρον


and him they found,

φρένα τερπόμενον


delighting his mind,

φόρμιγγι λιγείῃ||


with the clear-sounding lyre.

(Il 9.186)

The rising threefolder constitutes, as Kirk notes, a “substantial minority of Homeric verses,”55 and we could see it as a rhetorical strategy to alleviate the monotony of an indiscriminately prolonged series of periodic distichs. The rhetorical function of the tripartite hexametrical period would be a factor whether or not the participating cola have played a role in the origin of the hexameter. We also find tripartite sequences with other, less regular rhythmic profiles, e.g.:

a. ||εἰ δ’ ἄγε,


well then,

b. τούς ἂν ἐγὼ | ἐπιόψομαι,


whomever I will choose,

c. oἱ δὲ πιθέσθων. ||


let them obey.

(Il 9.167)

The central unit is longer than in the previous example, beginning earlier, right after the first dactylic foot, and running up to the bucolic diaeresis. In spite of the tripartite structure the middle caesura, falling after egṑ in unit b, is not absent and could be seen as slight break, rhetorical rather than cognitive, in the middle of unit b. The default rhythm of the hexameter, in other words, is not disrupted.56 The following example contains a tripartite hexameter that can be seen as a disruption of the basic rhythm of the period:

a. ||τὸν βάλε δεξιὸν ὦμον ·

him he hit in the right shoulder,

b. | ὁ δ’ ὕπτιος ἐν κονίῃσι

and he, on his back in the dust,

c. ||κάππεσεν οἰμώξας,

he fell with a cry,

d. | ἕταροι δέ μιν ἀμφὶ φόβηθεν

and his comrades around him they scattered,

e. ||Παίονες·

the Paionians,

f. ἐν γάρ Πάτροκλος

for in [them] Patroklos,

g. φόβον ἦκεν ἅπασιν

fear he sent into all of them,

h. ||ἡγεμόνα κτείνας,

having killed their leader,

i. | ὃς ἀριστεύεσκε μάχεσθαι. ||

who was the best of them in the fighting.

(Il. 16.289–92)

After two distichs (units a-b and c-d), the next verse is a tripartite period (units e-g) like the one in the previous example. But this time there is no rhythmic anticipation, as in the case of the distich or the rising threefolder. The middle unit f, with its five heavy syllables, seems out of place rhythmically, and would have been more appropriate as the first unit of a period.57 This metrical disturbance results from the insertion of unit e, a unit that is added to what precedes in the sense discussed earlier.58

The cognitive boundary between unit e and the preceding unit d is much less strong than that between unit e and the following unit f, and we might for that reason say that unit e is enjambing. Enjambement, commonly seen as the mismatch between a metrical unit (the hexameter verse) and a linguistic unit (the sentence), has long occupied an important place in the discussion of Homeric meter and style.59 In the perspective developed here, where sentence recedes in favor of speech unit as a cognitive entity and verse in favor of metrical period as a rhetorical one, the notion of enjambement might have to be modified accordingly. As we have seen in Chapters 4 and 5, the two major operations in Homeric syntax, continuation and framing, require a space that almost always exceeds the metrical period. To the extent that sentence is a phenomenon from a different medium, there-fore, I suggest that we refrain from using the term “enjambement” when-ever the progression of speech units is in accordance with the metrical period.60 This happens, for example, in the transitions from unit b to c and from unit g to h in the passage just cited, where we see a clause uttered within a frame and an adding participial phrase.61 In these cases the cognitive boundary between two speech units coincides with the rhetorical boundary between two periods.62

But there are varying degrees to which the movement from one focus of consciousness to the other does not coincide with the cyclic rhythmical movement of the period. Sometimes the boundary between two periods is only blurred, as it is between units d and e, and the disturbance that results is absorbed by unit f and contained within the metrical period. But some-times the cognitive units are at variance, not with the default rhythm within the period, but with the movement across two or more periods. For example, a unit may start at the bucolic diaeresis and move beyond the time that is left in the period (i.e., the clausula - ᴗ ᴗ - -) into the next period. The remarkable thing about such necessary or violent enjambement, with a strong mutual cohesion of the linguistic material on either side of the rhetorical boundary,63 is not that it occurs, but that it occurs in clusters, series of rhythmical mismatches creating tension that is sustained across two or more periods. To bring out the dynamic character of these rhetorically charged moments, we might speak of antimetry to characterize the secondary rhythm that is temporarily set up against the movement of the hex-ametric period.64 The second assembly of the Trojans provides an example:

a. || ἐς δ’ ἀγορὴν ἀγέροντο,

and they gathered for the assembly

b. | πάρος δόρποιο μέδεσθαι.

before thinking of food,

c. ||ὀρθῶν δ’ ἐσταότων

and with everybody standing

d. | ἀγορή γένετ’,

the assembly, it was held,

e. οὐδέ τις ἔτλη || ἔζεσθαι ·

and no one dared sit down,

f. πάντας | γὰρ ἔχε τρόμος,

for fear, it held them all,

g. οὕνεκ’ Ἁχιλλεύς || ἐξεφάνη,

since Achilles, he had appeared

h. δηρὸν δὲ | μάχης ἐπέπαυτ’ ἀλεγεινῆς. ||

and long he had stopped from the dire fighting.

(Il. 18.245–48)

The description begins with a distich (units a-b) and a further unit (c) that divides the time of the metrical period in the usual manner. The fourth unit (d), containing old information that is not cognitively salient (agorḕ génet’ ‘the assembly it was held’), serves as a bridge to unit e, where the antimetry begins; the unit runs into the next period, producing a limping, unhexametrical rhythm.65 Unit f then carries the sequence of ideas to the point at which the next antimetrical movement starts (unit g), at the bucolic diaeresis, exactly the same time in the metrical period as the previous one. Units e and g do not, of course, pretend to be “false starts,” hexameters beginning too early and thereby weakening the basic metrical cadence: no attempt has been made to conceal the beginning of the metrical cycle, as appears from the hiatus in unit e between étlē ‘dared’ and hézesthai ‘sit down’, and the virtual single-short rhythm due to the shortening of the long final vowel of étlē in line end. Indeed, the antimetrical effect of unit e and g derives precisely, not from their weakening the metrical cycle but from their acknowledging it; the units move against the basic rhythm, thus creating a tension that cannot but be rhetorically effective. This tension, we observe, is connected with an increase in semantic salience. Unit f, bridging the time between the two antimetrical units e and g, is as to its content also less prominent than the two surrounding units causing the metrical turbulence. This attraction of metrical (rhetorical) and semantic salience is even clearer in the next example:

a. αἰδοίης ἐκυρῆς ὀπὸς ἔκλυον,

of my honored mother-in-law I heard the voice,

b. ἐν δ’ ἐμοὶ αὐτῇ || στήθεσι πάλλεται ἦτορ

and within me the heart, it is pounding in my breast,

c. ἀνὰ στόμα,

up to my mouth,

d. νέρθε δὲ γοῦνα || πήγνυται ·

and my legs below, they cannot move,

e. ἐγγὺς δή τι κακὸν

yea, something terrible is close

f. Πριάμοιο τέκεσσιν.

to Priam’s children,

g. αἴ γὰρ ἀπ’ οὔατος εἴ η ἐμεῦ ἔπος ·

may the word it be far from my ear:

h. ἀλλὰ μάλ’ αἰνῶς || δείδω

but terribly I fear. …

(Il 22.451–55)

Andromakhe’s verbalization of her anxiety constitutes a highly functional case of antimetry: in three installments (units b, d, and h) she expresses her feelings, and each time her discourse breaks through the metrical structure, producing the halting, unhexametric rhythm of antimetry.66 One might want to consider such cases67 as a suppression of meter that is uncharacteristic of oral composition and not likely to occur in the practice of improvisation.68 Yet from the point of view of performance and audience attention—that is, rhetoric—one cannot help finding them very effective. They may even be indispensable, if not to underscore emotional or chaotic scenes by means of meter, then at least to avoid a metrically too correct and therefore potentially dull sequence of units.

Metrical versus Syntactic Periods

By way of apodosis to this chapter, we may add to this discussion of the emergence of meter from discourse that the unperiodic nature of Homeric syntax is by no means less complex than the periodic and hypotactic discourse of later periods. Rather, Homeric discourse is an equally complex result of the marking of ordinary speech into special speech, though of a completely different kind. As noted above, the essence of special speech lies in the connection between the discourse of the special occasion and ordinary discourse as the source on which it draws for its rhetorical effect. As such the concept of special discourse is useful in bringing out a common denominator of Homeric discourse and classical rhetorical prose, a similarity that gets lost in the more usual opposition between a formulaic Kunstsprache and the periodic Kunstprosa of later times.

Yet the concept is no less useful in bringing out the differences. Classical rhetoric, in increasing the importance of syntactic anticipation in speech and turning it into hypotaxis, can be said to have deviated from ordinary speech in an important step toward what we would call prose. By contrast, Homeric discourse has retained the characteristic flow of speech, but subjected it to the constraints of the hexametric period. It is this rhythmical period, the primary characteristic of the discourse of the performance, that turns Homeric speech into the stylization of ordinary speech. The flow of speech in the performance may be driven by the consciousness of the Homeric performer, but this consciousness, in its turn, is driven by the rhythmical flow of the hexametric period. Insofar as the audience cannot help participating, this process can rightly be called rhetorical.

1 See Chapter 3.

2 In fact, some authors argue that ordinary language is inherently poetic (Friedrich 1986: 24–27) or that ordinary language is not something that can be isolated from poetic language (Fish 1980: 97–111).

3 For a discussion of the differences between Aristotle’s use of the term περίοδος and later uses (Demetrius, Cicero), see Siebenborn 1987 (with more literature), who argues for an origin of the term in the sphere of dance and music. My metrical understanding of “period” in this chapter is in agreement with Siebenborn’s discussion.

4 Cf. ibid. 1408b28–30, another mention of the boundedness of number and the unpleasantness of the unbounded. The Greek terms ἀριθμός number’ and ῤυθμός ‘rhythm’ may have been connected by folk etymology; in Latin, a single term numerus is used (see below).

5 For detailed discussion of multiple constraints (including rhythm) in the composition and recall of epic and other oral genres, see Rubin 1995, the work of a cognitive psychologist. On rhythm and memorability see also Turner 1992: 93.

6 For discussion of rhetorical strategies in Homer, see also Mueller 1984: 11–13; Hainsworth 1993: 92–93 (focusing more on τάξις ‘structure, ordering’ than on λέξις ‘style’).

7 The concepts “rhetoric” and “rhetorical” are often (e.g., Cole 1990: 12; Ford 1992: 17) taken to imply professional reflection on language, and hence a distinction between form and content. My understanding of the concepts in this chapter is purposely wider and “pretheoretical,” covering any purposeful enhancement of speech in whatever situation. Rhetoric comes thus close to pragmatics; see Leech 1983: 15.

8 From Chafe 1990: 85, presented and analyzed again in Chafe 1994: 130–31 to illustrate “climax” in conversational narrative. In this method of transcription the pauses between intonation units are shown by dots: two dots indicate brief breaks up to one-half-second long and three dots mark longer pauses (up to one second). Numbers in brackets indicate measured pauses (in seconds); see Chafe 1994: xiii. The pauses within units g and k did not coincide with an intonational boundary in the speech recorded.

9 For a discussion of what I am calling the cognitive features of the passage in terms of activation, see Chafe 1990: 90–91. Notice the use of “and” which links all clausal intonation units to what precedes (see Chapter 4 on continuation).

10 I have been able to observe these physical features of this discourse, listening to the tape recording of which the cited text is a transcription. Notice also the pause within unit k, a rhetorical feature that does survive transcription.

11 See Chapter 5 above.

12 Note that if “totally” had to be placed in a standard English version of the same expression, it would fall exactly where the pause is in the middle of unit k.

13 For examples of more “professional” informal narratives, featuring both cognitive features (intonation units) and rhetorical features, see Polanyi 1982; Sobol 1992 on modern American storytelling; and especially the translation and transcription of Zuni Indian narrative in Tedlock 1972, who remarks (xix): “What makes written prose most unfit for representing spoken narrative is that it rolls on for whole paragraphs at a time without taking a breath: there is no silence in it.” He then goes on to explain his efforts to represent the breaks in Zuni narrative and to create a transcription that does justice to both the cognitive and rhetorical features of the original.

14 Rosenberg 1988, an investigation that started out, in a first edition, as an attempt to exemplify Parry’s and Lord s principles of oral composition on the basis of this African-American religious speech genre, but which later came to focus equally on the specific nature and expressivity of this type of discourse (1988: 4–5). On the performed African-American sermon, see also Davis 1985.

15 Alan Jabbour in Rosenberg 1988: 16–17, from a personal letter to Rosenberg. The description applies to the services of W. T. Ratliff in Durham, North Carolina in 1969, but according to Rosenberg fits over ninety percent of his material.

16 From Rosenberg 1988: 97–98, 222. The speaker is Rubin Lacy from Bakersfield, California. I have followed Rosenberg’s presentation, starting each line with a capital, and without Chafe’s prosodic punctuation.

17 Notice the Indo-Europeanists’ conception of apposition as a reduced independent clause (Ammann 1922; Schwyzer 1947). See Chapter 3.

18 On the coincidence of intonation units with rhythmic units in an oral tradition, see also Rubin 1995: 86.

19 Rosenberg 1988: 98; see also 1988: 90–91 on performer-audience interaction and metrical regularity.

20 On poetry “in” ordinary language, see also Sapir 1921: 221–31; Friedrich 1986: 24–27. On the traditional opposition between ordinary and poetic language, see Fish 1980: 97–111. These authors, however, do not speak of performance.

21 Cf. Turner 1992: 93–94: “Somehow the rhythm of the words is remembered even when the words themselves are lost to us; but the rhythm helps us to recover the mental state in which we first heard or read the poem, and then the gates of memory are opened and the words come to us at once.”

22 Cf., e.g., Kuipers 1993 on the epistemology of ritual performance; Chafe 1993 on the formulaic and prosodic differences between ordinary speech and the special speech evoking a remote source of authority. The neurophysiological aspects of performance, ritual, and rhythm are explored in d’Aquili and Laughlin 1979. In classical studies, see the work of Detienne (1967) and Vernant (1959) on truth and memory.

23 E.g., Rosenberg 1988: 28–29, 36–37.

24 On this opposition as the crucial difference between the Homeric and the South-Slavic tradition studied by Parry and Lord see Finkelberg 1990. On the Muses in general, see Ford 1992: 31–34, 52–53, 61, 72–76. On the eyewitness stance see Bakker 1993b; 1996a. See also Chapters 4 and 5.

25 Cf. Il. 2.485.

26 Translated in Chadwick and Chadwick 1932–40: 182; quoted by Finnegan 1977: 193, who is followed by Thalmann 1984: 224 and Dougherty 1991. Cf. Finkelberg 1990: 303. The phrase “springs from my inner self” could serve as a fairly accurate rendition of the Greek αύτοδίδακτος. Against the more usual interpretation of this word, “self-taught,” Fernàndez-Galiano objects (in Russo et al. 1992: 279–80) that “there would be little sense in Phemius boasting of being self-taught to Odysseus, a man who himself owes none of his skills to his teachers.” Ford (1992: 32) takes the phrase as implying independence from other poets and their work. It seems best to understand αύτοδίδακτος as “spontaneous,” referring to the production of poetry and song in the poet’s consciousness. See also Thalmann 1984: 127.

27 E.g., Lanata 1963: 13–14. Murray, who stresses the intellectual, nonecstatic character of the conception of poetic inspiration in Homer, comes to speak about the divine contribution in terms of knowledge and information (1981: 90–92), although she does object (97) to the form-content interpretation of Od. 22.347. On knowledge and the Muses, see Chapter 7 below.

28 See also Thalmann 1984: 126–27; Ford 1992: 32–33.

29 See also Dodds 1951: 10; Maehler 1963: 23.

30 See in particular Tannen’s discussion of conversational artistry, which ends with an analysis (1989: 173–95) of the chanted lines of Jesse Jackson’s address to the Democratic Convention of 1988.

31 Cf. Denniston 1952: 57, in connection with the artistic nature of the figure of speech referred to in ancient rhetoric and stylistics as hyperbaton, the marked use of a given word outside the syntactic environment to which it properly or logically belongs, and a subject usually treated under the general heading of word order: “The Greeks stylized everything; and it is the most difficult thing in the world to point to any Greek which may be regarded as ‘natural,’ ” the implication being that hyperbaton makes a speech unnatural and the equation of “artistic” with the latter. See also Norden 1909: 65–66 on rhythm as the cause of “unnatural” word order. Dover’s strategy, in his discussion of the word order problem in Greek syntax (1960), is to turn to inscriptions for basic and natural language; this attempt to escape the influence of Kunstprosa has to be understood in the same way. In the search for ordinary or natural language, the real problem might well be the difficulty of determining what ordinary language actually is (cf. Fish 1980: 97–111), rather than the shortage of data on nonliterary language.

32 In the example cited above, for example (“and the next minute, … [0.9] they just . . fell over… Totally”), the realization of “totally” as a separate unit is reminiscent of hyperbaton as a figure of speech (the “normal” word order being “the next minute they totally fell over”). The discussion of hyperbaton exclusively in terms of word order is typical of the stylistic study of a textual skeleton, not of the discourse itself. Could it be that some cases of deviant, hyperbaric word order in ancient texts are less-than-optimal recordings of passages in which syntactic separation is a consequence of intonational separation in the actual performance, recitation, or delivery of the text? See also the discussion of chiasmus and ring composition in Chapter 5 above.

33 Κῶλα or membra in ancient terminology. Aristotle (Rh. 1409b13–16) is the first to define κῶλον ‘limb’ as a constitutive part (“clause”) of a “period” (περίοδος). See also Dem. De eloc. 2–3. Quintilian (Inst. or. 9.123) defines the membrum (a Latin translation of Greek κῶλον) as a unit that is rhythmically complete but semantically meaningless when detached from the “body” of the period. One level below the membrum, Quintilian distinguishes the incisum ‘incision’ (Greek κόμμα), aptly called articula ‘joint’ by Cicero (e.g., De orat. 3.186), which he defines as a unit that is both semantically and rhythmically incomplete.

34 See Chafe (1994: 57), who cites Goldman Eisler 1968 for the relation between speech pauses and breathing.

35 Cic. De orat. 3.178–80. Cicero’s examples apply to the makeup of the universe, the organic unity of the human body and the structure of artifacts (columns of temples, for example, which are necessary for the solidity of a structure but add dignity to it as well).

36 Ibid. 175, 182, 184. Cf. also Arist. Rh. 1408b30–31: ῤυθμόν δεῖἒχειν τóν λόγον, μέτρον δέ μή· ποίημα γάρ ἒσται ‘A speech should have rhythm, not meter; otherwise it will become a poem’; Isocr. fr. 12: ὂλως δε ὁ λόγος μή λόγος ἒστω, ξηρόν γάρ· μηδέ ἒμμετρος, καταφανές γάρ· άλλα μεμείχθω παντί ῤυθμῷ ‘A speech should not be “speech” in its entirety, for that would be arid; but it should not be wholly metrical either, for that would be too obvious. A speech should display an even distribution of all sorts of rhythms’. Aristotle recommends the paean (rhythmic patterns of three short syllables and one long syllable) to begin and end periods (Rh. 1409a2–21). See also Siebenborn 1987: 233.

37 Cic. De orat. 3.186. Cf. Arist. Rh. I409b8–10.

38 Cf. Cole, who actually limits rhetoric as a concept (1991: 1–30) to this institutional sense and who claims that rhetoric in this sense did not exist before the fourth century (arguing partly on the basis of the absence of the term ῤητορική in fifth century texts).

39 See Cic., Orat. 175; Norden 1909: 41; Kennedy 1963: 68; Eisenhut 1974: 14.

40 Examples are antithesis: marked syntactic juxtaposition; isocolon: sameness of two or more cola; homoioteleuton: similarity in sound between the endings of various cola. See Norden 1909: 50–53; Kennedy 1963: 64.

41 See Cic., De orat. 3.173–74.

42 Cf. the remark of Gorgias την ποίησιν ἂπασαν νομίζω καί όνομάζω λόγον ἔχοντα μέτρον ‘All of poetry I consider to be and refer to as discourse with meter’, Hel. 9, a statement followed by a description of what cannot but be the psychology of the performance. Gorgias, then, seems to have been concerned with introducing prosaic features into poetry rather than with introducing rhythm and other poetic features into prose. To discuss Gorgianic discourse in terms of prose art, I submit, is to reduce his program to mere stylistic prescriptions pertaining to the properties of his speech as text, whereas he himself seems to have been more interested in his discourse as a way in which an orator can use, for his own ends and in his private interest, the effects of special speech in performance.

43 Norden 1909: 64, followed by a rhythmical analysis of other parts of Gorgias s speech.

44 See Meillet and Vendryes 1968: 598; Meillet 1937: 358–59·

45 See Chapter 4.

46 Though it would be misleading to characterize the speech genre of the folk sermon as being on its way toward more rhythmical regularity. On the marking of special speech as a diachronic process, see Nagy 1990a: 29–40.

47 This usage contrasts with those accounts that use the term “meter” for the abstract (ideal) profile of a given verse type, and “rhythm” for the concrete realizations of this type (e.g., Van Raalte 1986: 6; Sicking 1993: 43). I depart from such usage (though without questioning the validity of the observations articulated by it), in the same way that I prefer not to speak of realizations of a “formula” or a “sentence.”

48 Nagy 1974: 145; see also Nagy 1990a: 37. Useful formulations are also to be found in Devine and Stephens 1993: 399–400; 1994: 101. See also Allen 1973: 14.

49 On Rig-Veda see Kurylowicz 1970; Dunkel 1985: 119–20; Nagy 1990b: 31. For Old English see Cable 1974; 1991; Foley 1990: 110–20. For Kalevala see Schellbach-Kopra 1991: 135–36. On Serbo-Croatian see Lord 1960: 54–55; Foley 1990: 85–106. General remarks in Turner 1992: 61–105.

50 Some approaches treat cola as linguistic phenomena, with caesuras resulting from colon boundaries. See Frankel 1968: 100–156, for a four-colon theory. See also Porter 1951; Barnes 1986. Other accounts stress the structural, metrical properties of the verse. See for example Beekes 1972; Van Raalte 1986: 28–103; Sicking 1993: 70–71; Bakker 1988: 165–71. This approach may lead, whether or not explicitly, to a stance against historical and genetic accounts of the hexameter. See for example Hoekstra’s critique (1981: 3 3–53) of the genetic accounts proposed by Nagy 1974 and West 1973.

51 On the dactylic hexameter being longer than syntactic units (seen as formulas rather than as speech units) see Parry 1971: 191–239; Russo 1966; Ingalls 1972.

52 The origin of the hexameter is often thought of as the coalescence of a hemiepes -ᴗᴗ-ᴗᴗ- (X) and a paroemiac × - ᴗᴗ-ᴗᴗ--, reflected in the middle caesura (West 1973; 1982: 35; Haslam 1976: 202). Berg (1978), followed by Tichy (1981), argues for another coalescence with the Aeolic meter of the pherecrateus × × - ᴗ ᴗ - - - as second member, yielding what in hexameter metrics is called the hephthemimeral caesura as the genetically important joint. See also Gentili (1977; 1988: 15) who argues for an original looser association of the cola that were to produce the hexameter. The most elaborate discussion is Nagy’s (most recently 1990a: 459–64), who proposes a common ancestor for the hexameter, the Aeolic pherecrateus, and the dactylic cola of “dactylo-epitrite” meters.

53 In this representation, || signifies the beginning of a metrical period, and | the primary division of the period, the middle caesura of the verse.

54 See Kirk 1985: 20–21. Minton notes (1975: 33–34) that this type of division is more frequent in Hesiod than in Homer. In many cases (as perhaps in the example given) the first break is less marked or even very weak, so that a case could be made for a bipartite organization with a single break at the hephthemimeres.

55 Kirk 1985: 20.

56 Cf. also Il. 24.1: λΰτο δ’ άγών, | λαοί δέ | θοάς έπ’ι νήας ἒκαστοι || ἐσκίδναντ’ ἰέναι ‘and the games ended, | and [as for] the people, | to the swift ships each | they went dispersing’, where a slight break may perhaps be discerned after λαοί δέ, uttered as a separate unit before its clause.

57 Names like Patroklos (consisting of three long syllables) are much more at home either at the beginning of the line or before the middle caesura (see O’Neill 1942).

58 See Chapter 5. The present case is an example of disambiguation.

59 Since Parry’s 1929 article (1971: 251–65) enjambement has played an important role in discussions of oral composition. See, e.g., Lord 1960: 54; Kirk 1966; Edwards 1966; Bakker 1990b; Higbie 1990 (the fullest account); Clark 1994.

60 Cf. the earlier treatment of this topic in Bakker 1990b.

61 On the reasons for not viewing κάππεσεν (unit c) as the “verb” of a sentence of which ό is the “subject,” see Chapter 5.

62 This is what Parry (1971: 253) called unperiodic enjambement, the type of enjambement that is in accordance with the adding style (see also Chapter 3 above). Notice, however, that Parry would not consider the first enjambement, between units b and c, unperiodic; he would assign it, instead, to the more severe category of necessary enjambement (in which two necessary parts of a sentence are separated by the end of the line—on this point, see the previous note). For the various typologies of enjambement see Parry 1971: 253; Kirk 1976: 148; Higbie 1990: 29.

63 Some cases of adjective and noun separated by the end of the verse are Il. 1.78; 8.75, 128; 17.360, 371; 13.191; 24.122. Cf. also Edwards 1966: 125–33; Higbie 1990: 55–56, 115–16.

64 Edwards (1966: 136–37; 1991: 42–44) tends to stress the emphasis received by the word following the verse boundary in such cases.

65 The rhythm is unhexametrical because by Meyer’s First Law phrases of the form — or —ᴗ ᴗ - ᴗ almost never begin the verse, no doubt to avoid the very sequence with which it ends (Beekes 1972; Van Raalte 1986: 93; but cf. Kirk 1966: 97).

66 One might argue that unit b consists, after all, of two cognitive units separated by the end of the verse, but in that case the cohesion between them would be so strong that the effect produced would be in practice the same as described here. The same applies to units e and g in the previous example: one might want to argue that οϋνεκ’ Άχιλλεύς (unit g) is a unit on its own that frames ’εξεφάνη in the way described in Chapter 5. But that would not affect the analysis proposed here: repeated antimetrical effects would still occur, since the cohesion between ἔζεσθαι and έξεφάνη and the units preceding would be much stronger than between these verbal forms and the units following (f and h).

67 Cf. also Il. 6.407–11; 8.125–29; 10.149–54; 12.184–85; 13.687–92; 16.60–62, 107–10, 335–40, 367–68, 395–96, 552–53; 19–92–96; Od. 2.167–68.

68 See Kirk 1976: 168–69 on Il. 16.306–50.

Additional Information

Related ISBN
MARC Record
Launched on MUSE
Open Access
Creative Commons
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.