The Lyric Possession of an Epic Past
a machine readable edition
Copyright © 1980, 1997 The Johns Hopkins University Press. All rights reserved. This document may be used, with this notice included, for noncommercial purposes within a subscribed institution. No copies of this work may be distributed electronically outside of the subscribed institution, in whole or in part, without written permission from the JHU Press.
§1. The theory of oral poetry as set forth by Milman Parry and Albert Lord resists application to Archaic Greek poetry only if oral poetry is defined too narrowly by the opponents of the theory--and if the surviving poetry of Archaic Greece is treated too broadly as a general example of oral poetry. In what follows, I attempt not only to sketch a concept of oral poetry that is broad enough to accommodate the various forms of ancient Greek poetry but also to redefine these forms in terms of specific sub-types of oral poetry. 1
§2. The most secure basis for inquiry into the varieties of oral poetry is that of social anthropology. 1 From the vantage point of social anthropologists, various forms of song, poetry, and prose have functioned and continue to function in various ways in various societies without the aid of--in most cases without the existence of--writing. 2 From this vantage point we should not even be talking about oral poetry, for example, as distinct from poetry but rather about written poetry as possibly distinct from poetry: in other words, written poetry is the marked member of the opposition, and the poetry that we call oral is the unmarked. 3
§3. From the vantage point of our own times, however, poetry is by definition written poetry, and what we need to do first is to broaden our concept of poetry. Aside from questions of oral poetry and written poetry, the very word poetry becomes a source of confusion, in that it excludes dimensions normally included in the word song.
§4. The semantic differentiation between poetry and song affects the nomenclature of constituent elements common to these two differently perceived media: for example, whereas poetry is said to have meter, song has rhythm. This conventional distinction has a long history. There are traces attested in the scholarship of the fourth century B.C., where proponents of a rhythmical approach to poetry had an ongoing argument with proponents of a metrical approach. 1 The argument continues to this day, with the "metricians" emphasizing the patterns of alternating long and short syllables in the text as it is composed, and the "rhythmicians," the patterns of rhythm in the song as it is performed. In their argumentation the rhythmicians tend to define poetry in terms of song while the metricians tend to define song in terms of poetry. 2 My position is closer to that of the rhythmicians, to the extent that the affinities between song and poetry in ancient Greece can be viewed in terms of an evolution of various kinds of song into something differentiated from song--let us call it poetry--so that song and poetry can then coexist as alternative forms of expression. This point is elaborated as the discussion proceeds.
§5. Another point to be elaborated concerns Archaic Greek poetry and song in general, which I define for the moment as all attested poetry and song from Homer to Pindar: throughout this book, I argue against the need to assume that the medium of writing was necessary for the medium of composition or for the medium of performance and reperformance. 1
§6. Let us begin the extended discussion by considering the level of poetry, proceeding from there to the level of song. My premise stems from the observation of Albert Lord, based primarily on ethnographic field work in South Slavic traditions, that composition and performance in oral poetry are aspects of the same process, in that each performance is an act of recomposition. 1 Suffice it for now to add that so long as the traditions of oral poetry are alive in a given society, a written record cannot by itself affect a composition or a performance, and that it cannot stop the process of recomposition in performance. 2
§7. The basic forms of ancient Greek poetry are traditionally classified in terms of metrical types:
§8. In each of these metrical types of Greek poetry, I propose that the format of performance was recitative as opposed to melodic. This is not to say that such forms of poetry had no prescribed patterning in pitch. But patterns of pitch in poetry were formally and functionally distinct from the patterns of pitch that we, on the basis of our own cultural conditioning, recognize as melody in song. On the level of form, the difference is not as drastic as suggested by the contrast of monotone with song. I find the term recitative more suitable than monotone, to the extent that it does not necessarily convey the absence of melody. I use the term recitation to indicate either the absence or the reduction of melody. The contrast between not-sung (or recitative) and sung (or melodic) is attested most clearly in fifth-centry Athenian tragedy, where the iambic trimeter of dialogue was "spoken" by actors while a wide variety of other meters were sung and danced by a khoros 'chorus', to the accompaniment of an aulos 'reed'. 1 It bears emphasis that khoros 'chorus' in Greek is a group that sings and dances, to the accompaniment of wind or string instruments, and that, in Greek traditions, the concept of song is fundamentally connected with the concept of the chorus. 2
§9. In the claim just made for the iambic trimeter of Athenian tragedy, the argumentation is relatively secure. What follows, however, is a matter of controversy. I am proposing that an absence or at least a reduction of melody--and an absence of instrumental accompaniment and dance--eventually developed not only in the iambic trimeter of dialogue in Athenian drama but also in the iambic trimeter of the old iambic poets (Archilochus, Hipponax, Semonides, Solon, and so on), in the elegiac distich of the old elegiac poets (Archilochus, Callinus, Mimnermus, Tyrtaeus, Theognis, Solon, Xenophanes, and so on), and in the dactylic hexameter of Homer and Hesiod. This proposition may at first seem startling, in view of such internal testimony as Homer's bidding his Muse to sing the anger of Achilles (Iliad I 1) or Archilochus' boasting that he knows how to 'lead a choral performance' (verb exarkhô) of a dithyramb (F 120 W). 1 The significance of this evidence, however, is not what it may first appear, and we must examine it more closely. To begin, the internal evidence of Homeric and Hesiodic diction tells us that the word aeidô 'sing' (as in Iliad I 1) is a functional synonym, in contexts where the medium refers to its own performance, of the word e(n)nepô 'narrate, recite' (as in Odyssey i 1), which does not explicitly designate singing. 2 For some, the functional synonymity of aeidô 'sing' and e(n)nepô 'narrate, recite' is proof that the narrative format must be song--that the Homeric (and presumably Hesiodic) poems were sung and accompanied on the lyre. 3 For others, however, the equating of a word that refers to strategies of narrating Homeric and Hesiodic poetry with a word that refers to the format of singing to the accompaniment of a lyre proves only that such poetry had such a format in some phase of its evolution. 4 Self-references in Archaic Greek poetry may be diachronically valid without being synchronically true. 5 This phenomenon may be designated as diachronic skewing.
§10. For example, the epic poetry of Homer refers to epic poetry as a medium that was performed in the context of an evening's feast. Yet we know that the two epic poems of Homer, by virtue of their sheer length alone, defy this context. 1 If we look for the earliest historical evidence, we see that the actually attested context for performing the Iliad and Odyssey was already in the sixth century not simply the informal occasion of an evening's feast but rather the formal occasion of a festival of Panhellenic repute, such as the Panathenaia in Athens. 2 The performers at such festivals were rhapsôidoi 'rhapsodes'. 3 In Plato's Ion the rhapsode Ion is dramatized as just having arrived at Athens in order to compete in the rhapsodic contest of the Panathenaia (Ion 530ab). That the rhapsodes who performed at such festivals were in competition with each other is evident also from the reference in Herodotus (5.67.1) to agônes 'contests' (agônizesthai) in the public performance of "Homer's words" by rhapsôidoi 'rhapsodes' in the city-state of Sikyon, which were banned under the reign of the tyrant Kleisthenes. 4
§11. In the case of Homeric poetry, the earliest phases of rhapsodic transmission are associated with the Homêridai, a corporation of rhapsodes who traced themselves to an ancestor called Homêros (Pindar Nemean2.1; Plato Phaedrus 252b; Strabo 14.1.33-35 C 645; Contest of Homer and Hesiod, p. 226.13-15 Allen). 1 The basic testimony is most clearly set forth in the scholia to Pindar (Nemean2.1c, III 29.9-18 Drachmann), while the equation of the Homeridai with rhapsodes is specified in the actual text of Pindar (Nemean2.1-3).
§12. The scholia to Pindar (again Nemean2.1c, III 29.9-18 Drachmann) also specify that rhapsodes such as Kynaithos of Chios, credited with the final form of the Homeric Hymn to Apollo, could no longer trace themselves to Homer. In other words the tradition continued by Kynaithos is here being discredited by the sources as no longer authorized by the Homeridai. This Kynaithos is said (ibid.) to have been the first rhapsode to recite Homeric poetry at Syracuse, in the 69th Olympiad (504/1 B.C.), according to Hippostratus (FGH 568 F 5). By implication, Kynaithos was the first recorded winner in a seasonally recurring festival at Syracuse that featured a competition of rhapsodes. 1 As other possible examples of Homeric transmission not authorized by the Homeridai, I cite the traditions about the introduction of Homeric poetry to Sparta by Lycurgus the Lawgiver: he is said to have received the tradition not from the Homêridai of Chios but from the Kreôphûleioi (Kreophyleioi) of Samos, who traced themselves back to an ancestor called Kreophylos of Samos (Plutarch Life of Lycurgus 4). 2
§13. The ancient sources make it explicit that the rhapsodes, in performing Homeric poetry at the Panathenaia, were constrained by law to take turns in narrating the poetry in its proper sequence ("Plato" Hipparchus 228b and Diogenes Laertius 1.57). 1 In other words, even if the size of either the Iliad or the Odyssey ultimately defied performance by any one person at any one sitting, the monumental proportions of these compositions could evolve in a social context where the sequence of performance, and thereby the sequence of narrative, could be regulated, as in the case of the Panathenaia. 2
§14. From this rapid survey of rhapsodic traditions in the performance of Homeric poetry, I conclude that the model of simultaneous composition and performance by an oral poet at a feast had evolved organically into a quite different model, with the continuity of composed narrative achieved through a continuum of performance by rhapsodes who take turns at occasions like a Panhellenic festival. 1 The point that I am making about the context of Homeric performance applies also to the medium of performance: just as the Homeric testimony about the performance of epic by singers at feasts belies the synchronic reality of the performance of epic by rhapsodes at Panhellenic festivals, so also the Homeric testimony about the singer's singing to the accompaniment of the lyre belies the synchronic reality of the rhapsode's reciting without any accompaniment at all. On the basis of available evidence, it appears that rhapsodes did not sing the compositions that they performed but rather recited them without the accompaniment of the lyre. 2 So also with Hesiodic poetry: the internal testimony of the composition represents a theogony that is simultaneously sung and danced by the local Muses of Helikon (Theogony3-4, 8), 3 and yet we know that the Theogony, as also the other Hesiodic compositions, was in fact recited by rhapsodes. 4 This is not to say that hexameter could not be sung in the Archaic period: 5 only that hexameter evolved into poetry as distinct from song, and that its fundamental form of rendition, as poetry, was recitation. 6
§15. Similarly with old iambic and elegiac poetry we see that the internal testimony refers to choral singing and dancing to the accompaniment of the lyre (as in Theognis791; cf. 776-779), 1 or singing to both the lyre and the aulos 'reed' (531-534, 759-764), 2 or singing to the aulos alone (825-830, 943-944, 1055-1058, 1065-1068). 3 But in fact the external evidence points in another direction, namely, that the attested traditional format of performing the iambic trimeter and the elegiac distich in the Classical period and beyond was simply recitative. The crucial passage is Aristotle Poetics 1447b9-23, 4 about which it has been said that "it is our earliest explicit testimony about the mode of performance of elegiac." 5 This is not to say that elegiac distich and iambic trimeter had not in earlier stages been compatible with instrumental (and vocal) melody. In the case of elegiac we can cite not only the internal evidence of references to the aulos but even the possible external evidence of testimony in Pausanias and elsewhere for such earlier stages. 6
§16. The fact remains that the professional performers of old iambic and elegiac poetry were not singers but rhapsodes. 1 The crucial passages in this regard are Plato Ion 531a, 532a, with references to rhapsodic performance of the poetry of Archilochus, and Athenaeus 620cd, 632d, a difficult set of references to which we shall return presently. The notion of rhapsôidos 'rhapsode' can refer to amateur performances as well, as in Plato Timaeus 21b1-7, a passage that describes how the young Critias took part in "rhapsodic" contests in performing the poetry of Solon and others. 2
§17. All this is not to say that the references made in Archaic iambic or elegiac poetry to choral performance or instrumental accompaniment are diachronically wrong: as I have been arguing, they are in fact diachronically correct, 1 and it is not without reason that even the performance of a rhapsode is from a traditional point of view an act of "singing" (e.g., Plato Ion 535b). 2 Still, such references are synchronically inaccurate, becoming a source of confusion.
§18. In one particular case, the testimony of Athenaeus 620cd, 632d, 1 we must make a special effort to sort out the chronologically diverse strata of information. For example, Athenaeus 620c quotes Clearchus (F 92 Wehrli) as saying that one 'Simonides of Zakynthos used to perform, rhapsode-style, the compositions of Archilochus in theaters, while seated on a stool [diphros]' (ta Archilochou, phêsin, Simônidês ho Zakunthios en tois theatrois epi diphrou kathêmenos errapsôidei). 2 Also, there is a report of a rhapsode called Mnasion who performed the iambic poetry of Semonides (Athenaeus ibid.) 3 and of a rhapsode called Kleomenes who performed the Katharmoi of Empedocles at the Olympics (Dicaearchus F 87 Wehrli, by way of Athenaeus ibid. and Diogenes Laertius 8.63). 4 Similarly we read at Athenaeus 632d that the poetry of Xenophanes, Solon, Theognis, and the like was composed without melody (cf. also Aristoxenus F 92 Wehrli, with commentary). But we also read at Athenaeus 620c (= Chamaeleon F 28 Wehrli) that the poems of Homer, Hesiod, and Archilochus could be sung melodically. This statement follows up on the immediately preceding discussion, at Athenaeus 620b, of Homêristai. These Homêristai seem to be distinct from the rhapsôidoi, 5 and they represent the innovative practice of taking passages that were traditionally recited and setting these passages to music (cf. the references to Homer at Athenaeus 632d; there is also a similar reference to Hesiod in Plutarch Sympotic Questions 736e). Such activity was characteristic of the post-Classical era, 6 about which it has been said: "So great is the ascendancy of song over speech that, in the [Hellenistic] revivals of tragic and comic texts of the 5th and 4th centuries B.C., it even takes over the parts composed in iambic trimeters, intended originally for simple recitation." 7
§19. Once such a phase is reached, where traditionally recited pieces of poetry are being set to music, it becomes easy to confuse and reinterpret the diachrony of various sung and recited meters that are obviously related to each other. For example, apparently on the basis of parallelisms in meter and diction between Homer and Terpander, a representative of Archaic lyric who was credited with the composition of songs sung to the lyre, 1 Heraclides Ponticus is cited as saying that Terpander set his own poems and those of Homer to music (F 157 Wehrli in "Plutarch" On Music 1132c). So also Stesichorus is described as having set what are described as his epê, loosely to be translated as hexameters, to music ("Plutarch" On Music 1132c). As we see later, this hexameter of Stesichorus is cognate with, but not identical to, the Homeric hexameter. 2 Further, Archilochus is credited with the invention of the parakatalogê, that is, a delivery characterized by reduced rather than full melody, with the accompaniment of a musical instrument ("Plutarch" On Music 1141a). 3 In "Aristotle" Problems 19.6, the parakatalogê is described as a form of delivery that explicitly contrasts with song. 4 Archilochus is generally credited with 'introducing the practice of having some iambics spoken with instrumental accompaniment and others sung with it' (eti de tôn iambeiôn to ta men legesthai para tên krousin, ta de aidesthai, Archilochon phasi katadeixai "Plutarch" On Music 1141b).
§20. The reported distinctions, with reference to the "inventions" of Archilochus, between iambics that are sung as well as instrumentally accompanied and those that are not sung, that is, with reduced melody, though instrumentally accompanied, are valuable in revealing an intermediate stage in what I argue is an evolution from sung to spoken forms in the Archaic period. The claim that Archilochus was the actual inventor of the sung and the intermediate forms can be discounted as readily as the parallel claim that he invented the iambic trimeter (see On Music 1140f). What is essential is that these "inventions" are then correlated with historically attested innovations that start with the late fifth century, such as those of Crexus, in setting iambic meters to music (On Music 1141b); Crexus was a contemporary of Timotheus and Philoxenus (1135c). 1 In the post-Classical period, when poetry is being set to music--that is, the period starting with the likes of Crexus, Timotheus, and Philoxenus--we see that the archaic forms of song, as also of reduced song, like parakatalogê, are being treated as if they too were such "innovations". 2
§21. We can be satisfied with the diachronic correctness of ancient Greek poetry's references to itself as song by noting that such self-references are traditional, not innovative. The traditional phrases in Homeric poetry and elsewhere about the subject of singing and song have an Indo-European ancestry. 1 Even the word rhapsôidos 'rhapsode', designating the professional reciter of poetry, is built on a concept of artistic self-reference ('he who stitches together the song') that is likewise of Indo-European provenience. 2 The institutional reality of formal competition among rhapsodes, immortalized for us in Plato's dialogue Ion (530a), 3 seems to be a direct heritage of formal competition among singers, as reflected directly in passages like Homeric Hymn6.19-20 4 and indirectly in the numerous myths about such competitions. 5 There is enough evidence, then, to conclude that what the rhapsodes recited was directly descended from what earlier singers had sung. 6
§22. There is no compelling reason to believe that the medium of writing had anything to do with the traditions of the rhapsodes. 1 In fact there is positive evidence that their mnemonic techniques were independent of writing. The textual tradition of Homeric poetry as we have it stems from Hellenistic Alexandria, where the practice of accentual notation was invented. This textual tradition bears witness to certain archaic patterns of Homeric accentuation that were no longer current in the everyday Greek language--patterns that can now be verified through the application of Indo-European linguistics. 2 This comparative evidence leads to the conclusion that these patterns were preserved through norms of recitation inherited by the rhapsodes; the factor of writing seems to be ruled out, since a textual tradition for the notation of accents was evidently lacking before Alexandrian times. 3 Even in such matters of minute detail, we may infer that the oral tradition of the rhapsodes was inherited--albeit in an ossified or crystalized phase--from the oral tradition of the singers who came before them. 4
§23. Up to this point, I have used the term oral tradition only in a broad sense--to the extent that the medium of writing is not to be taken as a prerequisite for either composition or transmission. As we approach the subject of oral poetry in particular, I am for the moment more interested in the applicability of the term poetry, as distinct from song, to the oral traditions of ancient Greece. We have seen that a differentiation seems to have taken place, which can be represented in the following diachronic scheme: 1
By SPEECH I mean everyday or unmarked language, and by SONG I mean special or marked language that is set off from SPEECH on the formal level of phonology, morphology, syntax, or any combination of these three. 2 From a functional point of view, SONG would be any speech-act that is considered set apart from plain or everyday speech from the standpoint of a given society. 3
§24. The perception of plain or everyday SPEECH is a variable abstraction that depends on the concrete realization of whatever special speech, or SONG, is set apart for a special context. In small-scale societies, the setting apart would normally happen in terms of myth and ritual.
§25. I use the word ritual here not in terms of our own cultural preconceptions but in terms of the broadest possible anthropological perspective. 1 For the moment, I invoke the working definition of ritual offered by Walter Burkert: "Ritual, in its outward aspect, is a programme of demonstrative acts to be performed in set sequence and often at a set place and time--sacred insofar as every omission or deviation arouses deep anxiety and calls forth sanctions. As communication and social imprinting, ritual establishes and secures the solidarity of the closed group." 2 The insistence of ritual on a set order of things should not be misunderstood to mean that all rituals are static and that all aspects of rituals are rigid. Even when a given society deems a given ritual to be static and never changing, it may in fact be dynamic and ever changing, responding to the ever changing structure of the society that it articulates. Also, even within the strict framework of a given ritual, the various rigid patterns that conform to an ideology of unchangeability may be combined with various flexible patterns that conform to the needs of the here and now. Such is the case with the festivals of Greek city-states.
§26. As for myth, it can be defined for the moment as "a traditional narrative that is used as a designation of reality. Myth is applied narrative. Myth describes a meaningful and important reality that applies to the aggregate, going beyond the individual." 1
§27. In small-scale societies--rather than complex ones--we can observe most clearly the symbiosis of ritual and myth, how neither is to be derived from the other, and how the language of ritual and myth is marked, let us call it SONG, while everyday language, SPEECH, is unmarked. To repeat, the perception of plain or everyday speech is a variable abstraction that depends on the concrete realization of whatever special speech is set apart for a special context, let us call it occasion. In small-scale societies, the setting apart is normally a matter of ritual and myth, and the idea of ritual includes not only such basic activities as sacrifice and prayer but also such diverse occasions as meeting, eating and drinking, courtship, hunting, gathering, farming, building, and traveling. 1 The marked speech-acts associated with the special occasions of ritual and myth are what we are calling SONG. Internal criteria for marked speech-acts can be expected to vary from society to society: what may be marked in one may be unmarked or everyday in another. A striking example is the Bahutu convention of singing one kind of song while paddling upstream, another while paddling downstream. 2 There are potential differentiations of marked and unmarked categories within everyday language as well.
§28. Marked speech-acts could be a matter of denial as well as affirmation or reinforcement, as in the case of the Greek word an-ain-omai 'deny', opposed to an unattested *ain-omai 'affirm'. The form *ain-omai conveys that the thing affirmed is a social contract, in the broadest sense, as we can see from the derivative of *ain-omai, ainos, in the sense of 'legislative decision or resolution'. 1 In other words ainos is authoritative speech: it is an affirmation, a marked speech-act, made by and for a marked social group. 2
§29. In complex societies--and the situation in Archaic Greece can already be described as such--the pervasiveness of myth and ritual, as well as their connectedness with each other, may be considerably weakened. Still, the marking of speech, that is, the turning of unmarked SPEECH into marked SONG, may persist as the basic way to convey meaning in the context of ritual and myth. There is a reflex of this pattern in the usage of the Greek verb muô, which means 'I have my mouth closed' or 'I have my eyes closed' in everyday situations, 1 but 'I say in a special way' or 'I see in a special way' in marked situations of ritual. The latter meaning is evident in the derivatives mustês 'one who is initiated' and mustêrion 'that into which one is initiated, mystery [Latin mysterium]'. 2 So also the word mûthos 'myth', it has been argued, is a derivative of the same root from which muô is derived; 3 its special meaning seems to be 'special speech' as opposed to everyday speech. 4 For an illustration of the semantics underlying the usage of these Greek words, I cite Sophocles Oedipus at Colonus 1641-1644: the visualization and the verbalization of whatever it was that finally happened to Oedipus in the precinct of the Eumenides at Colonus are restricted, in that the precise location of his corpse is a sacred secret (1545-1546, 1761-1763). 5 Only Theseus, by virtue of being the proto-priest by hindsight for the Athenians of the here and now, is to witness what happened, which is called the drômena (1644). 6 Here the visualization and the verbalization of the myth, what happened to Oedipus, is restricted to the sacred context of ritual, controlled by the heritage of priestly authority from Theseus. 7
§30. From an anthropological standpoint, myth is indeed special speech in that it is a given society's way of affirming its own reality through narrative. 1 In Homeric diction, we see that the ancestor of our word myth, Greek mûthos, actually designates speech-acts, such as formal boasts, threats, laments, invectives, prophecies, prayers, and so on. 2 Let us for the moment take as a given, then, that the function of marked speech is to convey meaning in the context of ritual and myth.
§31. In most societies, not only the smaller-scale but the more complex as well, the pattern of opposition between marked and unmarked speech takes the form of an opposition between SONG and SPEECH respectively, with the "singing" of SONG being marked by a wide variety of patterns resulting from constraints on available features of SPEECH in the given language. From the standpoint of our own cultural preconceptions, singing is a patterning of both melody (stylized tone or intonation) 1 and rhythm (stylized duration and/or intensity). 2 From a cross cultural survey of a variety of societies, however, it is evident that singing may also be equated with many other available types of stylized phonological patterning, such as isosyllabism, rhyme, assonance, and alliteration, and that the patterning of SONG extends to the levels of morphology and syntax as well. 3 Moreover, there is a potential reinforcement of SONG with motor activity, as minimal as muscular tension 4 or as maximal as corresponding movement of the body in the form of dance. 5 With reference to dance, from this point onward, I argue that the activity of dancing to the words of SONG is primary, while dancing without the subtext, as it were, of SONG is secondary. 6
§32. This topic brings us to yet another type of markedness, yet another level of reinforcement for SONG, instrumental music. I also argue that instrumental accompaniment of the words of SONG is primary, while instrumental solo is secondary. 1 In the case of instrumental music, there can even come about a transition from marking speech as special to imitating special speech. 2 In making these arguments, my central point remains that the essential characteristic of SONG is the simple fact of its actual markedness or distinctness from everyday SPEECH.
§33. Let us pursue the argument with a tentative formulation on the level of phonology, considering the elements of rhythm and melody, prime constituents of our notion of singing. From the standpoint of the Greek language, what potentially sets SONG apart from SPEECH is a differentiation in patterns of duration / intensity (eventually rhythm) and pitch (eventually melody). 1 In a later stage of development what sets song apart from poetry is a further differentiation on the level of pitch (melody), so that song is plus melody while poetry is minus melody or reduced melody. The notion of plus melody is in line with such terms as lyric poetry or melic poetry, applicable to the medium of the main figure of this book, the poet Pindar.
§34. My view of poetry as something derived from SONG and differentiated from song runs counter to the view of metricians for whom song is poetry set to music. According to this second of two possible lines of thought, music would be extrinsic to language. This other view, however, runs counter to the experience of field work in ethnomusicology, a discipline that has built a strong case against the fallacy of treating music as a "universal language." 1 Our own cultural prejudices in favor of such a concept can be traced to medieval Europe, where the eventual dissociation of language and music was already under way. 2 Toward the end of the fourteenth century, Eustache Deschamps already made a distinction between the "natural music" of language and the "artificial music" of traditional melodies. 3 But it is clear in this case that the association of language and music is primary. For example, a study of attested traditions of Provençal singing has shown that only with the eventual divorce of melody from "text" can melody take on the characteristics that we, from the standpoint of our own cultural preconceptions, can recognize as music. 4 With the advent of polyphony, the motet can triumph over its libretto; but before that, in the twelfth and thirteenth centuries, the melodic traditions were still bound to phraseological traditions of song. 5 From the standpoint of medieval poetics, recognition as one good at melodies merely required a good vocal register; 6 a singer could be good at producing melodies and still be bad at producing words--and therefore a bad singer. 7
§35. Of the two terms, lyric and melic, the first is the more elusive in that it tends to be applied in contemporary academic usage to practically all Archaic Greek poetry except Homer and Hesiod. For my purposes, however, lyric is still the more useful term since it is more general. As such, lyric is suitable for distinguishing the general notion of song from the more specific one of poetry, which is restricted to the recitative medium of epic, elegiac, and iambic trimeter. From here on I use the word lyric as a parallel to song, excluding the elegiac and the iambic trimeter. In current usage such exclusion is generally not observed. It is instructive to notice, however, one particular constraint even in current usage against the application of the term lyric: we cannot say that the iambic trimeter of Athenian tragedy and comedy is lyric for the simple reason that it is patently recited as opposed to sung. As for what is sung, we call that lyric by way of opposition to what is recited. Thus the opposition of lyric meters and iambic trimeter in Athenian drama is that of song and poetry. We may note the dictum of Aristotle to the effect that iambic trimeter approximates, more closely than any other meter, everyday speech in real life (Poetics 1449a22-27; cf. Rhetoric 1408b33). Thus the opposition of song and poetry in tragedy not only recapitulates diachronically an earlier opposition of SONG and SPEECH: it also imitates synchronically the actual opposition of SONG and SPEECH in "real life."
§36. Whereas the iambic trimeter of tragedy regularly imitates SPEECH, the dactylic hexameter of epic occasionally imitates SONG, whenever Homeric poetry "quotes" a speech-act like a song of lament. 1 Thus the dactylic hexameter of epic, unlike the iambic trimeter of drama, is not used as a contrast to song; rather it can be used as an imitation of song. This distinction between these two kinds of poetry helps explain Aristotle's perception that the rhythms of dactylic hexameter, unlike those of iambic trimeter, are not close to those of everyday speech (Poetics 1449a27-28). In any case, the art of imitating SPEECH, as achieved in the medium of the iambic trimeter, can be measured by its realistic effects: the more the realism, the greater the artifice. 2
§37. Undifferentiated SONG as opposed to SPEECH can be imagined as having had features that ranged all the way from what we see in differentiated song to what we see in poetry. Thus, for example, SONG in any given society may or may not require melody. In other words what counts as poetry for us may in another given society count as song if there are no melodic prerequisites. In this light, I cite a particularly useful formulation by Dan Ben-Amos, based on a wide cross cultural variety of ethnographic data:
The existence or absence of metric substructure in a message is the quality first recognized in any communicative event and hence serves as the primary and most inclusive attribute for the categorization of oral tradition. Consequently, prose [= what I have been calling SPEECH] and poetry [= what I have been calling SONG] constitute a binary set in which the metric substructure is the crucial attribute that differentiates between these two major divisions. It serves as the definitive feature that polarizes any verbal communication and does not provide any possible intermediary positions. A message is either rhythmic or not. However, within the category of poetry [in my sense of SONG], speakers may be able to perceive several patterns of verbal metrical redundancy which they would recognize as qualitatively different genres. 1
§38. This statement, useful as it is, can be made more precise with reference to the term metric substructure. First, I turn to the cross cultural linguistic evidence assembled by W. S. Allen, showing that all phraseology has built-in rhythm. 1 In line with this thinking, I would argue that the inherited words of SONG contain the rhythm, from a diachronic point of view. In an earlier work on Greek and Indic metrics, I had put it this way:
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 non-traditional phraseology. By becoming a viable structure in its own right, meter may evolve independently of traditional phraseology. Recent metrical developments may even obliterate aspects of the selfsame traditional phraseology that had engendered them, if these aspects no longer match the meter. 2Such a formulation, to be sure, presupposes that the traditional phraseology of SONG, generating fixed rhythmical patterns, is itself already regulated by principles of phonological, morphological, and syntactical parallelism and repetition that serve to differentiate SONG from SPEECH. 3
§39. Granted, a factor like rhythm may become stylized to the point that it can become transferred from the words of song, in the forms of dance and instrumental music. 1 I am arguing, however, that patterns of convergence and parallelism between the rhythm in the words of SONG and the rhythm in the forms of dance and instrumental music are primary, while patterns of divergence and contrast are secondary. In any case the perception of rhythm depends ultimately on the innate human capacity for language. 2 In that sense the very process of dance is related to the inner rhythms of language. The linguistic factor of stress, which seems to be the basis of rhythm in languages like Greek, is psychologically perceived in terms of body movement. 3 We may compare the discussion in Plato Laws 653e-654a (also Laws 665a), where the combination of rhythmic and melodic idiom is synthetically visualized as khoreiâ 'choral song and dance'. 4 I am proposing that the phraseology of SONG tends to stylize and regularize its own built-in rhythms, and that these regularizations result diachronically in what we call meter. 5
§40. Similarly, I also propose that the phraseology of SONG can stylize and regularize its own built-in tones or intonations, resulting diachronically in what we call melody. 1 If we combine the two proposals, we get a scheme where both rhythm and melody in SONG could be viewed as regularized outgrowths of SPEECH that serve eventually to distinguish SONG from SPEECH. In terms of this composite scheme, I am now ready to substitute rhythm or melody or both for metric substructure in the formulation of Ben-Amos. The result is a formulation that is not alien to ethnomusicology. On the general topic of the connections discovered by ethnomusicologists between music (what I have been calling SONG) and language, I cite a general theory, built on a broad cross cultural sampling of ethnographic data, offered by Bruno Nettl:
My own theory is based on the assumption that an undifferentiated method of communication existed in remote times, one which was neither speech nor music [= SONG] but which possessed the three features that they hold in common: pitch, stress [which I reinterpret in the specialized sense of intensity], and duration.... There must have been a long, gradual stage of differentiation and specialization in culture, during which the two [= language and "music"] became distinct.... This theory, then, postulates three stages in the development of music: (1) undifferentiated communication, (2) differentiation between language and music, and (3) differentiation between various musical styles. The last stage is, of course, the only one for which we have any data at all, and even that...is fairly recent. 2
§41. In this connection, consider the earlier theories of Curt Sachs, 1 postulating three kinds of origins for melodic traditions: "logogenic" (from language), "pathogenic" (from motion), and "melogenic" (from "music"). On the basis of what we have noted about the relationship of language and motor activity, the category "pathogenic" is unnecessary. As for the category "melogenic," it may be useful for describing historically attested situations where a given melodic tradition has lost or at least outgrown its "logogenic" moorings, and where such a tradition is then recombined with or superimposed on originally unrelated phraseology. 2 Still, I would offer a formulation for melody that parallels what I have offered for rhythm: that the primary situation is that of convergence and parallelism between the patterns of tone or intonation or both in the words of SONG on the one hand and the patterns of melody, dance, and instrumental accompaniment on the other hand. 3 I would also argue, conversely, that the secondary pattern is that of divergence and contrast. 4
§42. It should be stressed, however, that contrastive patterns between dance or instrumental accompaniment on the one hand and song on the other, even if they are diachronically secondary, are even more effective than parallel patterns in marking off the language of SONG from the language of SPEECH. Intensified contrast in form further marks what is already marked in function. We should expect partial contrast, for example, in the patterns of melody in the song and of tone or intonation in the words of the song, or in the patterns of ictus in the verse and of stress (duration and intensity) in the words of the verse, or in the patterns of the colon in the stanza and of the clause or phrase in the words of the stanza, and so on. 1
§43. In light of these arguments, supported by the insights of ethnomusicologists, I offer a broadened outline of possible developments, with special reference to the development of Greek music. Whereas SONG may or may not have required melody, song must be plus melody as opposed to poetry, which is minus melody or reduced melody. Whereas SONG may or may not have required dance and instrumental accompaniment, given forms of song may be plus dance or plus instrumental accompaniment or plus both.
§44. Let us pursue further the point, made earlier, to the effect that the parallelisms between patterns of dance or instrumental accompaniment and patterns of rhythm or melody in SONG are diachronically primary and that the contrasts between them are secondary. 1 If indeed SONG is marked SPEECH, then such elements as dance and instrumental accompaniment can be viewed as ramifications of SONG that can in turn be further differentiated as either parallel to the SONG or contrasting with it or, even further, parting with it altogether, as in forms of dance or instrumental music that exist independent of SONG. This is not to say something altogether naive and pseudo-historical, such as "in the beginning there was song, which was both danced and instrumentally accompanied." Rather it is to speak of the linguistic foundations of singing, dancing, and instrumental accompaniment. It is to speak of diachronic potential: SONG, as a marked form of language, is structurally capable of generating differentiated subforms such as dance and instrumental music. From a diachronic point of view, then, dance and instrumental music are optional realizations of the stylized speech act. From the standpoint of traditions with song, dance, and instrumental accompaniment surviving together, analogous forms with any of these consituents missing are liable to be viewed as the result of a tearing away of that constituent from a unified whole, as we read in Plato Laws 669d-670a. In this connection, we may follow the formulation of A. M. Dale, who makes use of Milton's concept of Voice and Verse as uniting to form Song: "For the Greek lyric poet Voice and Verse were not a pair of sirens; Verse was merely the incomplete record of a single creation, Song." 2
§45. To set up language as the diachronic foundation of dance and instrumental music is in line with A. M. Dale's view that "song, with its dance, was a function of the words themselves when they were alive--that is, in performance." 1 More fundamentally, it is in line with Aristotle's view that the basis of musical rhythm is the syllable (Metaphysics 1087b33 and following). 2 Still, the fundamental function of dance and instrumental music, whether their patterns are parallel or contrastive with the patterns of language that they accompany, is to mark special speech as opposed to everyday speech, that is, SONG as opposed to SPEECH. 3 An ideal example is Athenian drama, in which the dancing and instrumental accompaniment further distances the words sung in the lyric meters by the khoros 'chorus' from the words recited in the iambic trimeter by the actors.
§46. Let us examine more closely the medium of the chorus, as attested in Athenian drama, in which song, dance, and instrumental accompaniment survive together. To repeat, analogous forms that happen to lack any of these consituents are viewed as a tearing away of that constituent from a unified whole (Plato Laws 669d- 670a). Wherever song has the capacity of being danced, as in the case of Greek choral lyric, dancing to the song is dancing to its rhythms and melodies on the level of form and to its words on the level of content. In Greek, this correspondence is mîmêsis or mimesis, which can best be translated as 'reenactment' or 'impersonation'. 1
§47. In general the noun mîmêsis, as well as the corresponding verb mîmeisthai, designates the reenactment, through ritual, of the events of myth (e.g., Lysias 6.51). 1 In the case of a highly stylized ritual complex like Athenian drama, the act of reenactment, mîmeisthai, is equivalent to acting out the role of a mythical figure (e.g., Aristophanes Women at the Thesmophoria850). 2 The acting out can take place on the level of speech alone or else on the level of speech combined with bodily movement, that is, dance: hence the force of pros 'corresponding to' in the expression pros ta pathea autou 'corresponding to his sufferings [= pathos plural]' at Herodotus 5.67.5, describing the singing and dancing by tragikoi khoroi 'tragic choruses', at Sikyon in the time of the tyrant Kleisthenes, in reenactment of the pathea 'sufferings' of the hero Adrastos. 3
§48. While the fundamental meaning inherent in mîmêsis is that of reenacting the events of myth in ritual, 1 by extension mîmêsis can also designate the present reenacting of previous reenactments. This narrowed view of reenactment, where the focus is on the the present reenactment of a previous reenactment without considering the whole chain of reenactments extending from the past into the future, corresponds to the more specialized and episodic notion of imitation. In that the newest instance of reenacting has as its model, cumulatively, all the older instances of performing the myth as well as the "original" instance of the myth, mîmêsis is a current imitation of earlier reenactments. This is the sense of mîmêsis in the Homeric Hymn to Apollo163, where a choral group called the Deliades are described as being able to 'imitate',' mîmeisthai, the voices and musical sounds of a wide variety of Ionians who are described as assembling for a festival on the island of Delos (162-164). 2
§49. The sense of wonder about the mîmêsis performed by the Deliades concerns the accuracy or exactness of their re- enactment: everyone will say, when they hear the sound of their own voices reenacted by the Deliades, that they are hearing their own way of speaking (Hymn to Apollo163-164). 1 This line of thought corresponds to the celebrated description of mîmêsis in the Poetics of Aristotle as the mental process of identifying the representing 'this' with the represented 'that': houtos ekeinos 'this is that' (1448b17). 2 In the performance of the Deliades, the represented 'that' is not only whatever the visiting Ionians have sung before. Whatever they have sung before is simply the latest in an ongoing series of previous reenactments, ultimately reenacting a given myth. So also with the formulation of mîmêsis by Aristotle (again Poetics 1448b17): the represented 'that' identified with the representing 'this' can be perceived not only as the previous experience but also as the sum total of previous experiences. 'This', then, is particular, the experience in the here and now, whereas 'that' is potentially universal, a cumulative synthesis of all previous experience. Aristotle goes on to say that the mental process whereby 'this' is being identified with 'that', by way of mîmêsis, is a source of pleasure (Poetics 1448b11, 13, 18). This pleasure is not incompatible with an anthropological understanding of ritual: "Fixed rhythm and fixed pitch are conducive to the performance of joint social activity. Indeed, those who resist yielding to this constraining influence are likely to suffer from a marked unpleasant restlessness. In comparison, the experience of constraint of a peculiar kind acting upon the collaborator induces in him, when he yields to it, the pleasure of self-surrender. 3
§50. Such a formula of equating the particular 'this' with the universal 'that', as implied by the use of the verb mîmeisthai in the Homeric Hymn to Apollo and as explicitly linked with the concept of mîmêsis in the Poetics of Aristotle, is a fundamental expression of assent in many languages: besides the many attestations in Greek (e.g., tout' ekein' 'this is that' = 'yes' in Aristophanes Lysistrata240), there is a particularly striking example in the Latin combination hoc illud 'this is that', which is the ancestor of the French oui 'yes' (and of the Southern French oc, as in Languedoc). Such a yes can serve as the amen of a participant in a given ritual, who assents to the realities of myth as reenacted in the context of ritual. 1
§51. From the standpoint of ritual, then, the activity of the chorus in an institution like Athenian drama, where, song, dance, and instrumental accompaniment can function as a unified whole, is a matter primarily of reenactment, insofar as the performers reenact the events of myth, and only secondarily of imitation, insofar as the performers at one given occasion imitate the performances of previous occasions.
§52. This is not to say, however, that reenactment is not imitation. All ritual reenactment is imitation, though of course not all imitation is ritual reenactment. Moreover, imitation is pertinent to the differentiation of SONG into song as opposed to poetry: we see a synchronic use of such a differentiation in Athenian drama, where the opposition between sung or lyric meters on one hand and the spoken iambic trimeter on the other hand imitates the real-life opposition of SONG on the one hand, with its ritual context, and SPEECH on the other, with its nonritual context. The imitation is effective: poetry actually seems closer than song to SPEECH in that it does not have the same degree of specialized patterning in melody. Nor is it correlated with dance or in most cases with instrumental accompaniment. 1
§53. And yet, if indeed poetry is to be derived from SONG, it is really one step further removed from SPEECH: to repeat the diachronic construct, song is specialized by retaining and refining melody from SONG, while poetry is specialized by losing or failing to develop the melody that is potential in SONG. 1 In terms of differentiation, some form of SONG had to lose melody, or fail to develop melody, so that poetry could be differentiated from song. In Athenian drama, this form was the iambic trimeter. From a diachronic point of view, however, this meter did not have to be the form that imitated SPEECH: we hear from Aristotle (Poetics 1449a21) that the trochaic tetrameter catalectic had been the earlier format of spoken poetry as opposed to song. 2 But the conventions of Athenian tragedy seemed to allow only one meter to serve as the canonical format for imitating SPEECH at any one given time: 3 in attested tragedy, for example, the trochaic tetrameter catalectic is not isofunctional with the iambic trimeter--it is marked off from it by virtue of being associated with "scenes of heightened tension." 4 Moreover, there is evidence that the trochaic tetrameter catalectic was in certain situations delivered in a reduced melodic form known as parakatalogê. 5 There is no need to argue, however, that iambic trimeter could never be sung after having become the imitative format of SPEECH: there are sporadic traces, even in Athenian drama, of sung iambic trimeter 6 as also of sung dactylic hexameter 7 and sung elegiac distich. 8 Still, the appropriate way to imitate the single format of SPEECH with the multiple formats of SONG is to contrast a single spoken meter with the plurality of sung meters. If Aristotle Poetics 1449a21 is right in saying that the trochaic tetrameter catalectic used to be the medium for imitating SPEECH, then I am ready to posit a stage where even iambic trimeter, like the trochaic tetrameter catalectic, used to be delivered in the format of parakatalogê, and where this type of modified melodic delivery used to be the only approximation of SPEECH. 9 Then another stage of differentiation could have led to the iambic trimeter of Classical tragedy, with its non-melodic delivery, while trochaic tetrameter catalectic persisted with a modified melodic delivery. At such a stage of differentiation, only iambic trimeter could imitate SPEECH, whereas the trochaic tetrameter catalectic would be imitating something that is now more than just SPEECH.
§54. At another stage further removed, the appropriate way to imitate the single format of SPEECH with the multiple formats of SONG would be to contrast a non-metrical form with the plurality of metrical forms. The nonmetrical form would be prose:
I suggest that the opposition between recited meters on one hand and spoken prose on the other hand once again imitates the real-life opposition of SONG vs. SPEECH. Again, the imitation is effective: prose seems closer than poetry to SPEECH in that it does not have the same degree of specialized patterning in rhythm. And yet, if indeed prose is predicated on poetry, as Herodotus implies in the first sentence of his Histories, 1 then prose is really one step further removed from SPEECH: to extend the diachronic construct, while song is specialized by retaining and refining melody from SONG, poetry is specialized by detaching melody from SONG, and prose is specialized by at least partially detaching rhythm from poetry. 2 Further, just as one form of poetry can coexist and interact with many forms of song in the medium of Athenian Drama, so also the form of prose coexists and interacts with forms of poetry in such forms of expression as represented by the Lives of the Seven Sages tradition and even by the Histories of Herodotus. 3
§55. Before proceeding any further in our consideration of the distinctions between song and poetry, we must confront a semantic problem occasioned by our own cultural preconceptions. Whereas the stylized rhythms of poetry are known to us as meter, we think of the stylized rhythms of song simply as rhythm. This mode of nomenclature is hardly appropriate to the traditions of Greek lyric, where song operates on principles of rhythm that are clearly cognate with the principles of meter in the recitative poetry of, say, Greek epic. In fact it is common practice to speak of the rhythms of Greek lyric in terms of meter.
§56. This much said, I should note that my theory about poetry as a differentiated derivative of SONG can be supported by the ancient Greek metrical evidence. To make this point, I offer in the Appendix a survey of two fundamental types of Greek lyric metrical patterns, the so-called Aeolic and dactylo-epitrite meters. All the attested lyric poetry of Pindar, with only a few exceptions, is composed in one or the other of these two kinds of meters. 1 In the Introduction, I singled out Pindar's compositions as the centerpiece for illustrating the traditions of lyric poetry, that is, song. In later chapters we see that the actual content of this poetry reaffirms both explicitly and implicitly the thematic connection between epic and lyric, which reflects on the formal connection between poetry and song. In the Appendix, finally, I present the case for arguing that these connections are verified by the form--or, better, forms--of lyric poetry.
§57. Specifically, I argue that the dactylic hexameter, the meter of epic, took shape from the phraseology of the two basic metrical systems inherited by Pindar, the Aeolic and the dactylo-epitrite (both surveyed at length in the Appendix). More generally, I also argue that the three major types of meter in Greek poetry, that is, dactylic hexameter, elegiac distich, and iambic trimeter, are differentiated equivalents of cognate types of meter found in Greek song, as in the inherited medium of Pindar. Whereas elegiac distich and iambic trimeter can be connected with dactylo-epitrite, dactylic hexameter is more easily explained in terms of both the Aeolic and the dactylo-epitrite heritage of Pindar. 1
§58. Further, it can be shown that the three basic meters of poetry, dactylic hexameter, elegiac distich, and iambic trimeter, are not only derived from SONG but also differentiated from the corresponding meters of song. To begin, we may note that the meters of Greek song are either strophic, built on the principle of the colon, the constitutive element of both the relatively simpler stanzas of monody and the more complex stanzas of the choral medium, or stichic, built on the principle of the verse. 1 Other available terms of description are asynartetic, where the divisions between cola are overt, and synartetic, where they are latent. 2
§59. The notion of asynartetic is compatible with that of parakatalogê, a category of song where both melody and dance become reduced or eliminated altogether. 1 Whereas both melody and dance are bound to a strophic framework, the category of parakatalogê is associated with an asynartetic or stichic framework. 2
§60. The stichic meters of lyric are a transitional point of differentiation from song, whereas total differentiation is achieved in the stichic meters of nonlyric, that is, of poetry proper. 1 The stichic rather than strophic meters of lyric are actually attested as usable for extended narrative of a type parallel to epic, composed in the dactylic hexameter, which is the stichic meter of nonlyric par excellence. A worthy example is Sappho F 44 V, a poem with a heroic setting: it is composed in a stichic meter 2 that is clearly cognate with various strophic meters of Lesbian lyric poetry. 3 Such stichic meters of narrative lyric poetry, conventionally sung to the accompaniment of the lyre, are doubtless more closely related than is the spoken meter of hexameter to the format of the South Slavic guslar who sings to the accompaniment of the gusle. 4
§61. Let us reformulate in terms of the concepts just introduced. The fundamental argument in the Appendix is that the sequences of cola in the strophic meters of song are cognate with the verses in the stichic meters of poetry. Perhaps the clearest example is the case of the strophic sequences of Stesichorus, which are built from units such as the following: 1
|B" m_oo_oo__||= pros~|
These shapes are not only prototypical of those found in, say, the so-called dactylo-epitrite meters of a Pindaric strophe: they are also identical with some of the major components in the meters of poetry, that is, in the dactylic hexameter (CB"), in the elegiac distich (CB"|CC), and in the iambic trimeter (bM). 2
§62. These meters of poetry are not only derived from SONG but also differentiated from the corresponding meters of song. As a finishing touch of differentiation between poetry and song, meters of song can avoid patterns that have been appropriated for poetry: thus, for example, the dactylo-epitrite metrical repertoire of Pindar's compositions contains the ingredients needed to generate equivalents of the dactylic hexameter, elegiac distich, and iambic trimeter, and yet it is precisely these patterns of spoken poetry that the lyric poetry of Pindar, let us call it song, consistently avoids. 1 It appears that the parent forms were covering their tracks. The poetic structure of the parent forms shades over, within that structure, those of its aspects that match various aspects that have become distinctive features of the respective daughter forms.
§63. With any differentiation of poetry from song through the loss of melody, there would have to come about a new structural strain in the oral tradition. Melody can be an important feature in the mnemonics of oral tradition in song, as we know from the studies of folklorists who scrutinize the transmission and diffusion of song: melody helps recall the words. 1 We are reminded of the anecdote about the Athenians captured after the debacle at Syracuse who ingratiated themselves with their captors by singing passages from Euripides: these memorable passages were evidently parts from choral lyric, not iambic trimeter (as we see from the wording tôn melôn aisantes 'singing from his lyrics' in Plutarch Nicias 29.3). 2 In terms of a differentiation of oral SONG into oral poetry as opposed to oral song, 3 I offer this axiom: with the structural strain brought about by the loss of melody in poetry, there would come about, for the sake of mnemonic efficiency, a compensatory tightening up of rules in the poetic tradition. 4 This tightening up would entail an intensification of both phraseological and prosodic regularities, as we see in the formulas and meters of Homer, Hesiod, and the old elegiac and iambic poets. I also suggest that the concept of formula, stemming ultimately from Milman Parry's study of Homeric phraseology, applies primarily to such regularities stemming from the differentiation of oral poetry from oral song. In other words the formula is to be seen as characteristic primarily of oral poetry as opposed to song. In order to account for the distinct regularities of oral song as opposed to poetry, the concept of formula could be considerably broadened. 5
§64. For song, it seems most useful to distinguish three dimensions of regularity: phraseological, rhythmical, and melodic. These three dimensions correspond to the tripartion of melos 'lyric poetry' into logos, rhuthmos, and harmoniâ in Plato Republic 398d (cf. Aristotle Poetics 1447a21-23). The rhythmical dimension would be represented by meters and the melodic, by modes. 1 The process of oral composition in song, then, can be conceived as an interaction of phraseology, rhythms, and modes.
§2n1. For my methodology, a particularly influential work has been Jacopin 1981, with its balanced treatment of parole as well as langue. Cf. also Leach 1982, especially p. 5, with incisive comments on the impact of Jacopin's work.
§2n2. For a forceful presentation, with an emphasis on oral song and poetry, see Zumthor 1983. At p. 34 the author stresses that oral poetry is not poetry minus writing. As an introduction to the characteristics of oral poetry, the standard works remain and will surely remain Parry  and Lord 1960. The intellectual and emotional resistance to the findings of Parry and Lord stems for the most part from various cultural preconceptions of our own times concerning "folk poetry"; for an illuminating historical account of such preconceptions, centering on the dichotomy of "Volkspoesie" and "Hochpoesie," see Bausinger 1980.41-55 ("Folklore und gesunkenes Kulturgut"). Cf. also Nettl 1965.13: songs can travel not only from "high" culture to "low," but also the other way around. In the case of German traditions the two-way travel between "art music" and "folk music" is particularly intense (Nettl, p. 69). As songs travel "up" and "down," there can be commensurate patterns of tightening or loosening, either way, in the built-in rules of song making.
§2n3. Cf. Intro. §16.
§4n1. Cf. Pöhlmann 1960.29-48, especially 29-32, 47-48; also Henderson and Wulstan 1973.48-49 on the different systems of notation used by the rhythmicians and metricians.
§4n2. Cf. Allen 1973.96-125, especially pp. 98-100.
§5n1. On the role of the written text as an alternative to performance, see the discussion in Ch.6, especially with reference to the work of Svenbro 1988. My argument that writing is not essential for either the composition or the performance of poetry and song in the Archaic period of Greece requires, already at this point, one major modification: a notable exception is the Archaic epigram, which does indeed require the medium of writing as an alternative to performance, though not for composition. As Alexiou 1974.13 and 106 argues, the epigram is a poetic form that compensates for emerging patterns of restriction against antiphonal types of lamentation performed in two choral subdivisions, where one subdivision took the role of the dead, engaging in a "dialogue" with another subdivision that took the role of the living; the medium of writing was necessitated as a substitute for actual performance, in the wake of social pressures, exerted within the new context of the emergent city- states, against ostentatious degrees of lamentation on the part of families with powerful ties to the older phases of the existing social system. Even in the case of the epigram, it can be argued that writing had no direct role in the actual composition of the poetry: it appears that the built-in mechanics of composition, which can be ascertained from the diction of the various attested epigrams, do not necessarily correspond to the various local patterns of spelling reflected by these epigrams. Two notable examples in Archaic epigrams are (1) the spelling-out of elided vowels (e.g., CEG 13.4) and (2) the spelling of "movable n" in violation of the meter (e.g., CEG 288; cf. Kock 1910.22). For an internal cross reference to the genre of the epigram within Homeric poetry, cf. Iliad VII 89-90 and the commentary (with bibliography) of Gentili and Giannini 1977.22-25. As for the various other forms of song and poetry, which were not dependent on writing as their primary vehicle of expression, I agree with the general arguments of Herington 1985.41-57 (especially pp. 46-47) against presupposing the necessity of writing as an aid for the performance of songs in what he calls the "song culture" of the Archaic and Classical periods. I disagree, however, with his postulating the necessity of writing for the actual compositon of songs (especially p. 41).
§6n1. Cf. Lord 1960.13-29, 99-123; cf. Nettl 1983.247-258 on the concept of field work.
§6n2. Cf. also Ch.2§5, Ch.2§13, Ch.2§24n6, Ch.3§4; Ch.13§2, Ch.13§48 and following.
§7n1. For the moment, I shall include under the rubric "Homer" not only the Iliad and Odyssey but also the Homeric Hymns and the poems of the Epic Cycle, such as the Aithiopis and Destruction of Ilion attributed to Arctinus of Miletus (Proclus, p. 105.21-22 and p. 107.16-17 Allen, Suda s.v.), the Little Iliad attributed to Lesches of Mytilene (Proclus, p. 106.19-20; Phaenias F 33 Wehrli, in Clement Stromateis 1.131.6), and so on. I reserve for Ch.2§38 a discussion of the patterns of differentiation between Homeric and Cyclic Epic. As we see in that discussion, as also later in Ch.14, the patterns of attribution to Homer become progressively more exclusive as we move forward in time, from the Archaic to the Classical period and beyond.
§7n2. I use the term wisdom poetry to encompass both the Theogony and the Works and Days.
§8n1. As the discussion proceeds, we shall see that some types of meter that are performed by the chorus are transitional between not-sung and sung, such as the so-called parakatalogê, with reduced rather than full melody (Ch.1§19) and with reduced dancing (Ch.1§53n9).
§8n2. Details in Ch.12, where I also reckon with various lines of argumentation that have been invoked to challenge the notion of an inherited correlation of song and dance in the khoros.
§9n1. The meter in which this utterance is composed is trochaic tetrameter catalectic, on which see Ch.1§54, Ch.1§53, Ch.1§53n2; also Ch.13§30. Also in the same meter is Archilochus F 121 W, where the description 'leading the choral performance' (again, verb exarkhô) applies to the choral leader of a paean. Further discussion of the concepts of dithyramb, paean, and choral performance (verb exarkhô) in Ch.3 and Ch.12. For more on Archilochus F 120 W, see N 1979.252n.
§9n2. Thus for example the aoidê 'song' of the Muses at Hesiod Theogony104 is in the context of the poet's bidding them to 'narrate' (espete: Theogony114) and to 'say' (eipate: Theogony115). On ennepô as 'recite', see N 1974.11n29.
§9n3. See for example West 1981, who makes this additional observation at p. 113: "We cannot make a distinction between two styles of performance, one characterized as aeidein, the other as enepein."
§9n4. Again, N 1974.11n29.
§9n5. I am using the terms diachronic and synchronic, on which see Intro. §11n1, not as synonyms for historical and current respectively. It is a mistake to equate diachronic with historical, as is often done. Diachrony refers to the potential for evolution in a structure. History is not restricted to phenomena that are structurally predictable.
§10n1. For further exploration of this subject, see N 1979.18-20. Note Kirk's (1962.281) comparison of the size of the Homeric compositions with the "leap from the largeish pot to the perfectly colossal one" in the evolution of monumental amphoras / craters during the Geometric Period. What interests me in this comparison is that the colossal size of a utensil defies its own utility (N, p. 20 §5n5).
§10n2. For a convenient collection of testimonia concerning the performance of Homeric poetry at the Panathenaia, see Allen 1924.226-227: Lycurgus Against Leokrates 102 (the law requires the performance of the poetry of Homer at the Panathenaia, to the exclusion of other poets), Isocrates Panegyricus 159, "Plato" Hipparchus 228b, Diogenes Laertius 1.57 (Life of Solon). Cf. also Hesychius s.v. Braurôniois. Herington 1985.139 calculates that the running time of the Oresteia trilogy of Aeschylus "could be more than a quarter of that of a full-length Homeric epic." He concludes: "These and similar figures seem often to be overlooked in discussions about the practicability of delivering the Homeric epics complete on any one occasion" (p. 269n58). A dynasty of tyrants in Athens, the Peisistratidai, played a major role both in the shaping of the Panathenaia and in making this festival the context for performance of epic (scholia to Aristides Panathenaicus 3.123; "Plato" Hipparchus 228b). The involvement of the Peisistratidai in the institutionalization of Homeric performance at Athens has been explained in terms of a "Peisistratean Recension" (for an introduction to this concept, with bibliography, see [S.] West 1988.36-40). The present book develops an alternative explanation that does not require the textual notion of a "recension." For more on the Peisistratidai and their connection with the performance of Homeric poetry at the Panathenaia, see especially Ch.2§40n3, Ch.2§44n3, Ch.2§49n5, Ch.6§28, Ch.6§53, Ch.6§83n3.
§10n3. For example, "Plato" Hipparchus 228b, concerning the rhapsôidoi 'rhapsodes' at the Panathenaia. It appears that some cities were later than others in instituting formal occasions for rhapsodic performance: see Maximus of Tyre 17.5a concerning the "latecomers" Sparta, Crete, and Cyrene (opse gar kai hê Spartê rhapsôidei, opse de kai hê Krêtê, opse de kai to Dôrikon en Libuêi genos).
§10n4. On agôn as 'contest' in poetry, see Homeric Hymn6.19-20 (cf. Ch.5§2n7). When Heraclitus (22 B 42 DK) says that Homer and Archilochus should be banned from agônes 'contests' in poetic performance, what is really being said is that rhapsôidoi 'rhapsodes' (as suggested by the playful use of rhapizesthai) should not be allowed to perform Homer and Archilochus. The expression homêreiôn epeôn 'Homer's words' in Herodotus 5.67.1 probably refers to the Seven against Thebes tradition, not to the Iliad or Odyssey; see Cingano 1985, whose argumentation meshes with a line of thought that pervades this book: that the patterns of attribution to Homer become increasingly less exclusive as we move further back in time.
§11n1. Cf. Brelich 1958.320-321. Elsewhere I have argued that the "signature" in the Homeric Hymn to Apollo172, where the speaker refers to himself as 'the blind man of Chios', is an idealized self-reference to Homer: see N 1979.5 and 8-9 (for a similar conclusion, from a different point of view, see Burkert 1979.57); the verb oikei 'he has an abode' (from noun oikos 'house, abode') at line 172 suggests that Homer, as ancestor of the Homeridai, had a hero cult at Chios (cf. N 1985.76-77, 81 §79n1). For other references to the Homeridai of Chios, see Acusilaus FGH 2 F 2, Hellanicus FGH 4 F 20 (both by way of Harpocration s.v.); Isocrates Helen 65; Plato Republic 599d, Ion 530c.
§12n1. For a defense of the reported date, 504/1 B.C., cf. West 1975, Burkert 1979, and Janko 1982.261-262n88; Burkert adduces, for comparison, a tripod with an epigram dedicated by Simonides on the occasion of his victory in a dithyrambic competition at Athens in 476 B.C. (Simonides EG 28). The relative lateness of the date here assigned to Kynaithos is puzzling to those who posit a relatively early date for Stesichorus, supposedly the earliest attested poet in the Hellenic West. But note the juxtaposition of Homer and Stesichorus in, for example, Simonides PMG 564 (Burkert 1979.56n16); also in Isocrates Helen 64-65. I interpret such references to imply the appropriateness of conventionally juxtaposing performances of Homeric and Stesichorean compositions at a given festival.
§12n2. Further details at Ch.2§42n2. Cf. Burkert 1972, who offers an analysis of myths that connect the transmission of Homeric poetry by the Kreophyleioi with the transmission of Homeric poetry by the Homeridai. This testimony about the reception of Homeric poetry at Sparta may be compared with the remark of Maximus of Tyre 17.5a, cited at Ch.1§10n3.
§13n1. This detail about taking turns is apparently not taken into account in the arguments of Schnapp-Gourbeillon 1982.720 against N 1979.18-20 (et passim).
§13n2. It remains to ask whether, in the case of the Panathenaia, the reported law about consecutive recitation (cf. also Ch.1§10) was a reinforcement or extension of something that might already have been a convention of, say, the Homeridai. Cf. also Lycurgus Against Leokrates 106-107 about a customary law at Sparta concerning the performance of the poetry of Tyrtaeus.
§14n1. Cf. Ch.1§10. I infer that the rhapsodes who took turns reciting within the sequence were in competition with each other. I cite again the reference in Herodotus (5.67.1) to agônes 'contests' (agônizesthai) in the public performance of "Homer's words" by rhapsôidoi 'rhapsodes' (cf. Ch.1§10n4). As H. Pelliccia suggests to me, the requirement for consecutive performance by rhapsodes has the effect of ensuring that competition does not result in the arbitrary selection, by ambitious rhapsodes, of the most popular sequences. [S.] West 1988.39-40 leaves room for the possibility that the division of the Iliad (and, by extension, of the Odyssey) into twenty-four "books" reflects traditional units of performance by rhapsodes at the Panathenaia. On the usage of rhapsôidiâ in the sense of a 'book' of the Iliad, see, e.g., Plutarch Apophthegmata 186e.
§14n2. The expression alura 'without lyre' in Plato Laws 810bc furnishes explicit testimony (cf. also Plato Ion 533b5-7). For testimonia about reciting rhapsodes holding a staff instead of a lyre, see West 1966.163-164 (though I disagree with his application of these testimonia to Hesiod Theogony30). The iconographic evidence of vase paintings showing rhapsodes either with a lyre or with a staff (West ibid.) can be viewed as another example of the phenomenon that I have called the diachronic skewing of perspective on an evolving institution (on which see Ch.1§9).
§14n3. See Ch.12 on the diachronic correctness of the description, in Hesiod Theogony3-4, 8, of song and dance in the performance of the Muses.
§14n4. For testimonia on the rhapsodic recitation of Hesiodic poetry, see Plato Ion 531a, 532a, Laws 658d; also Ch.1§22n4. For an overview of the evolution from singer (aoidos) to reciter (rhapsôidos), see N 1982.43-49. Conversely, the concept of rhapsode can be retrojected all the way back to Homer and Hesiod, as when Plato refers to both as rhapsodes (Republic 600d).
§14n5. For example, the hexameters attributed to Terpander, which counted as a lyric form, were sung: "Plutarch" On Music 1132c (Heraclides Ponticus F 157 Wehrli) and the commentary of Barker 1984.208n18. On the lyric hexameters (and quasi-hexameters) attributed to the archaic figure called Terpander as a model for those of the post-Classical poet Timotheus (including his attested Persians): On Music 1132de and the commentary of Barker, p. 209n25.
§14n6. See, for example, Aristotle Poetics 1447a29-b8, 1448a11, 1449b29; Plato Laws 669d-670a. Cf. Else 1957.37-39.
§15n1. Cf. Ch.12§55.
§15n2. Cf. Ch.12§55n5.
§15n3. Ibid. See also Archilochus F 58.12 W, where we read aidôn hup' aulêtêros 'singing to the accompaniment of the aulos- player'; cf. Theognis533, 825. Note Hipponax F 153 W and Mimnermus T 5 GP, by way of "Plutarch" On Music 1134a: the author is discussing an unattested passage of Hipponax where the poet cross refers to Mimnermus as an aulos-player (cf. T 2 and T 4 GP); this cross reference leads the author to assert that in earlier times elegiac poetry was sung to the accompaniment of the aulos. See also the report in Pausanias 10.7.5-6 that at an early stage elegiac distichs were performed to the accompaniment of the aulos at the Pythian Games, and that this practice came to an end in 582 B.C. Also, the testimony of "Plutarch" On Music 1134a suggests the possibility that at an early stage elegiac distichs were performed to the accompaniment of the aulos at the Panathenaia. In the case of Pausanias, Bowie 1986.23 argues that the specification of elegeia 'elegiac distichs' may have resulted from a misunderstanding of the word elegoi in the Echembrotus epigram quoted by Pausanias (Echembrotus, p. 42 GP II).
§15n4. See the commentary of Else 1957.56-57, who integrates the evidence of this passage with that of Poetics 1459b32-1460a2 and Rhetoric 1409a7; cf. also Lucas 1968.61.
§15n5. Rosenmeyer 1968.218 (by "elegiac" he is referring to the metrically determined category of elegiac distich). Rosenmeyer's 1968 article (following Campbell 1964) concentrates on the elegiac distich because his purpose is to challenge the widespread view, encouraged by the internal testimony sketched in the preceding discussion, that elegiac distich was regularly accompanied by the aulos. As for iambic trimeter, it is more generally agreed that it was recited, not sung (cf., e.g., Gentili 1985.45-46).
§15n6. Cf. Ch.1§15n3.
§16n1. I suggest that the repertoire of rhapsodes would include such lengthy compositions in elegiac couplets as the Smyrneis of Mimnermus, on which see Bowie 1986.27-30.
§16n2. Cf. Brisson 1982.61.
§17n1. See Ch.1§15. This formulation expands on the positions taken by Campbell 1964 and Rosenmeyer 1968. It can also serve as a friendly amendment to Bowie 1986.27, with whose basic point about the compatibility of the aulos 'reed' and the elegiac distich I agree.
§17n2. On this point, see the survey by West 1981.114n8 on attestations of aeidô 'sing' as designating the performance of a rhapsode. Cf. also Timaeus 21b1-7, as cited immediately above. I agree with Campbell 1964.66 that the expression en ôidêi 'in song [ôidê]' in Plutarch Solon 8.2 (with reference to Solon F 1 W) refers to "formal recitation like that of a rhapsode." A similar point can be made about the use of âidô 'sing' in Philochorus FGH 328 F 216 (by way of Athenaeus 630f). These considerations affect the arguments of Bowie 1986.19n29.
§18n1. Cited at Ch.1§16.
§18n2. Cf. West 1981.125.
§18n3. For the reading "Semonides" instead of "Simonides," see West ibid.
§18n4. West 1981.125 dates these testimonia to the fourth century B.C., or the end of the fifth at the earliest. On the setting for the performance of the rhapsode Kleomenes, we may compare the report that Dionysius I of Syracuse engaged rhapsodes to perform his poetry at Olympia (Diodorus Siculus 14.109).
§18n5. See the useful references of West 1970.919.
§18n6. As we see from a survey by Gentili 1979.26-31, with a focus on the performance of drama. Cf. also West 1986 on the hexameters of a newly-discovered inscription from Epidaurus, a fragment of a hymn that he dates "not later than the third century B.C." (p. 45; cf. p. 44n19). The melodic notations preserved in this inscription reflect, in West's opinion, a uniform instrumental cadence, hexameter after hexameter.
§18n7. Gentili, p. 26.
§19n1. See Ch.1§14n5.
§19n2. Cf. Appendix §30.
§19n3. Cf. the commentary of Barker 1984.212nn183, 185.
§19n4. Cf. Aristides Quintilianus, pp. 5.25-6.7 (ed. Winnington-Ingram) on the recitation of poetry as an intermediate category, to be placed between the categories of speech and song. See further at Ch.1§40n1. Cf. also Barker, p. 234n183.
§20n1. Cf. the comments of Barker, p. 52n20.
§20n2. For a similar line of thinking, note the report of Timomachus (FGH 754 F 1 in Athenaeus 638a), who says that one Stesandros was the first to set Homer to the lyre for a performance at Delphi; it is as if the medium of "Homer" had never been sung before, only spoken.
§21n1. Cf. N 1974.10n29 and 244-261.
§21n2. Durante 1976.177-179. The notion of rhapsôidos 'rhapsode' as 'he who stitches together the song' is made explicit in Pindar Nemean2.1-3. On the concept of oimê as a sort of textual "fil d' Ariane," see Svenbro 1976.45n135.
§21n3. Cf. also Herodotus 5.67.1, on which see also Ch.1§10n4. For further testimonia from inscriptions recording various contests of rhapsodes, see West 1981.114n13. Cf. also Brisson 1982.62-63, with a convenient summary of details, gleaned from the Ion of Plato, about the competition of rhapsodes at the Panathenaia.
§21n4. Quoted at Ch.2§47n3. Further details on this passage at Ch.5§2n7.
§21n5. For a survey of the institution of competition among singers, see Dunkel 1979; cf. N 1979.311 §2n6. For an example of a myth about such a competition, I cite the story of a contest between Arctinus of Miletus and Lesches of Mytilene, two of the poets of the Epic Cycle (Phaenias F 33 Wehrli, in Clement Stromateis 1.131.6). A more famous example is the Contest of Homer and Hesiod tradition (pp. 225-238 Allen); for bibliography, see Janko 1982.259-260n80; cf. also Dunkel 1979.252-253. On the interrelationship of narrative structure between the Homeric Iliad and Odyssey on the one hand and the Epic Cycle on the other, see Ch.2§38.
§21n6. For further arguments, see N 1982.43-49.
§22n1. This is not to say that in historical times they could not have owned texts of what they recited (cf. Xenophon Memorabilia 4.2.10); in any case, it is clear that the rhapsodes recited from memory (Xenophon Symposium 3.6).
§22n2. N 1982.45 and 69, citing Wackernagel  p. 1103; also West 1981.114n12.
§22n4. Note also the bits of information adduced by Allen 1924.48 about the sunthutai Mousôn Hêsiodeiôn 'fellow-sacrificers to the Hesiodic Muses' (IG VII 1785; cf. also 4240), a corporation that "owned the land at Thespiae which contained the sacred spots": Allen offers the theory that this corporation was analogous to the Homêridai in that it seems to have exercised authority over the corpus of Hesiodic poetry. At p. 72 of Allen's book, we find a parallel, not adduced at p. 48, that can serve as a powerful additional argument in favor of Allen's theory: in Plutarch Banquet of the Seven Sages 149f-150a, there is mention of one Ardalos of Trozen, supposedly a contemporary of the Seven Sages and described as both an aulôidos 'aulos-singer' (aulos = 'reed') and a hiereus tôn Ardaleiôn Mousôn 'priest of the Ardalean Muses'. The cult of these "Ardalean Muses" had been supposedly established by the ancestor of this Ardalos, also called Ardalos of Trozen (Plutarch 150a), who is elsewhere reported to be the inventor of the aulos (Pausanias 2.31.3). This parallelism suggests that the 'fellow-sacrificers to the Hesiodic Muses' are rhapsôidoi who transmit the compositions of Hesiod, just as the 'priests of the Ardalean Muses' are aulôidoi who transmit the compositions of their ancestor Ardalos--and just as the Homêridai 'sons of Homer' transmit the compositions of their ancestor Homer.
§23n1. I print song types, not just song, to indicate the potential plurality of song types in opposition with any single given type of poetry. I elaborate on this point in what follows.
§23n2. On the distinction between unmarked and marked members of an opposition, see Intro. §12.
§23n3. On the notion of speech-act, see Intro. §17.
§25n1. Cf. Leach 1982, especially pp. 5-6. For further elaboration, see Ch.4§3. Most Classicists of my generation tend to resist the very concept of ritual in the wake of the earlier excesses of the so-called Cambridge School: see, for example, Herington 1985.123-124. Herington's important contributions to our understanding of the earlier forms of Athenian tragedy could be further enhanced through a broader perspective of ritual. In this regard I find it helpful to cite the sketch offered by Seaford 1984.10-16. I agree with Seaford that his findings help confirm "the unfashionable view that the performance of tragedy originated in the practice of ritual" (p. 14).
§25n2. Burkert 1985.8.
§26n1. My translation, with slight modifications, of Burkert 1979b.29. For an illuminating discussion of myth, especially useful to those who are unfamiliar with the perspectives of social anthropology, I cite Leach 1982.
§27n1. Cf. Leach, pp. 5-6.
§27n2. Nettl 1965.120.
§28n1. For example, SIG 672.15. Commentary and further discussion by Edmunds 1985.105.
§28n2. Further details at Ch.6§4.
§29n1. Chantraine DELG 728: from onomatopoeic mû, with the primary meaning of opening and closing the lips (cf. Aristophanes Knights 10: see Chantraine, p. 717).
§29n2. We may compare the semiotics of whispering, which may count as a form of not-speaking in everyday or unmarked situations and as a form of special speaking in marked situations of secrecy, sacredness, and the like.
§29n3. See Chantraine, pp. 718-717, with bibliography in support of the argument that mûthos 'myth' is likewise derived from onomatopoeic mû. Despite the morphological grounds for accepting this derivation, Chantraine expresses doubts on semantic grounds. I hope that my interpretation here helps dispel that doubt.
§29n4. For more on mûthos 'myth', see Ch.2§28 and following. "Surviving examples" in social institutions tend to reflect a general pattern in earlier stages but only a particular situation in later stages of attestation.
§29n5. For the belief that the corpse of the hero was a talisman of fertility for his native or adoptive community, see Ch.6§59. On the hero cult of Oedipus at Colonus as represented by the tragedy of that name, see Edmunds 1981, especially p. 223n8 (a reference that is accidentally omitted in N 1985.76-77).
§29n6. The participle drômena is from the verb draô, which means 'do, perform' within the the world of tragedy but also 'sacrifice, perform ritual' within the "real world," the outer world that frames the world of tragedy. See Ch.13§13. The participle drômena, as used outside of tragedy, designates 'ritual' (e.g., Pausanias 9.27.2, 9.30.12; cf. Burkert 1983.33n14). Inside tragedy, as here at Sophocles Oedipus at Colonus 1644, it bears the more inclusive and ambiguous sense of 'things that are done', 'things that are happening'.
§29n7. N 1982b. The key lines are in Sophocles Oedipus at Colonus 1641-1644.
§30n1. On the truth-value of myth: Leach 1982.2-7.
§30n2. Detailed demonstration in Martin 1989.12-42. On the concept of speech-act, see Intro. §17.
§31n1. When differences in pitch have a lexical function (as in ancient Greek), it is a matter of tone; where they have a syntactical function (as in English), it is a matter of intonation: cf. Devine and Stephens 1985.151.
§31n2. On duration and intensity as aspects of "stress," cf. Devine and Stephens, p. 152. Further discussion at Ch.1§33n1.
§31n3. See Guillén 1985.93-121, especially pp. 103-104, and Bright 1963.29. One feature of the fusion of experience in ritual, as Tambiah 1985.165 suggests, is "the hyper-regular surface structure of ritual language: the poetic devices such as rhyme, meter, assonance, and alliteration generate an overall quality of union and a blurring of grammatical boundaries." Cf. also Jakobson 1960.358.
§31n4. Note the following remark of Merriam 1964.119: "Some connection is made between pitch and muscle tension; the musician becomes accustomed to the muscle tension which he knows to be correct. One Basongye musician expressed this by saying that he chooses a pitch 'which does not make me sweat', and the same musician very logically noted that he comes to know the voices of the people with whom he sings and thus chooses a starting pitch 'in the middle' which he knows will suit all the voices."
§31n5. Cf. Allen 1973.100. The concept of dance should not be defined narrowly on the basis of our own cultural preconceptions. The categories of stylized bodily movement corresponding to our notion of dance vary from society to society. Cf. Royce 1977. Further details are at Ch.1§39n3. On dance as an optional element in ballad performance, see Nettl 1965.56; for more details, with reference to Faroese culture, see Wylie and Margolin 1981.99, 115, 117. I note in particular the following description, p. 99:
At the village dancehall--or, before villages had dancehalls, in a house rented for the occasion--men and women link arms to form a long, twisting circle. Anyone may join the circle at any point. They dance with a rhythmically shuffling, kicking step to the singing of the ballads. There is no instrumental accompaniment. A skipari (leader) sings the verses of a ballad, while the rest of the singers join in on the verses (if they know them) and on the refrain. When one ballad ends, the ring keeps moving round for a few moments until a new skipari starts up a new one.
§31n6. Cf. the ethnographic testimony discussed briefly by Merriam 1964.275.
§32n1. Cf. Herzog 1934, Schneider 1957.32-33, and especially Sultan 1988.396-397. Note too Bake 1957.196-197 on the Indic traditional teaching that vocal music is "pure" sound while instrumental music is a "manifestation" of sound. As Nettl 1965.51 points out, the limitations of the human voice (not to mention the limitations of the human ear), as contrasted with the relatively greater freedom of sound-range in musical instruments, lead to differences in the patterns of evolution for vocal and instrumental music. In this connection it is useful to ponder the discussion of Bright 1963.27. See in general the survey of the relationships between language and music in Nettl 1964.281-292. On the tendency of specialization and even professionalization in the social position of those who perform SONG with instrumental accompaniment and, by extension, of those who perform musical instruments, see Nettl 1965.50. On the development of instrumental solo playing on the aulos, so that the aulôidos 'he who sings to the accompaniment of the aulos [reed]' gives way to the aulêtês 'he who plays the aulos', see Pausanias 10.7.4, with the terminus of 582 B.C. at the Pythian Games.
§32n2. Hence the notion of "talking instruments," as discussed by Stern 1957; cf. also Ong 1977b. On instrumental music as imitation of the "special speech" of bird song, see Merriam 1964.75. Conversely, at one step further removed, unusual vocal techniques like Alpine yodeling can be traced back to the imitation of instruments: cf. Nettl 1956.58.
§33n1. On "stress" in ancient Greek, which includes the phonological features of duration and intensity but not pitch, see the fundamental work of Allen 1973; for an updated defense of Allen's formulation, see Devine and Stephens 1985. From the standpoint of general phonetics, stress may be a matter of duration, intensity, and pitch. From a survey of typological evidence, Devine and Stephens, p. 152, point to "instances of languages in which intensity is independent of both pitch and duration (Japanese), languages in which intensity is independent of duration and combines with pitch as an exponent of stress (Estonian, Komi), and languages in which intensity combines with both pitch and duration as an exponent of stress (English)." When differences in pitch have a lexical function, as in ancient Greek, it is a matter of tone; where they have a syntactical function, as in English, it is a matter of intonation: see Ch.1§31n1. In ancient Greek, pitch is thus a matter of morphology as well as phonology. This is being taught today as the sum total of Greek accentuation. Allen's discovery, that ancient Greek also had a system of duration and intensity that was independent of its system of pitch, suggests that the two systems merged in Modern Greek, where the inherited patterns of pitch are correlated with both duration and intensity (Devine and Stephens, p. 146n83).
§34n1. Cf. Merriam 1964.10-11.
§34n2. See Zumthor 1972.100.
§34n4. Zink 1972.24: "Quand le divorce entre le texte et la musique sera comsommé, la musique, paradoxalement, pourra prendre plus d'importance; elle sera développée pour elle-même et pour l\'effet extérieur qu'elle produit, indépendamment des exigences internes du poème."
§34n5. Zink, pp. 17-24; especially p. 23n2.
§34n6. Examples in Zink, p. 23n1.
§34n7. Zink, p. 20n3.
§36n1. A prominent example is Iliad XXIV 723-776, where the narrative gives a direct quotation of three different laments, performed by three of Hektor's female next of kin on the occasion of the hero's funeral. At this funeral there are also professional aoidoi 'singers' (XXIV 720) who sing a more stylized kind of lament, called the thrênos (721), while the nonprofessional singers, next of kin to the deceased, are singing a less differentiated kind of lament, called the goos (XXIV 723, 747, 761). Correspondingly, at the funeral of Achilles, his next of kin, the Nereids, sing undifferentiated laments (Odyssey xxiv 58-59), while the Muses sing a differentiated thrênos (xxiv 60-61). Cf. N 1979.112.
§36n2. On the concept of imitation as a narrowed version of the concept of reenactment, see Ch.1§48.
§37n1. Ben-Amos 1976.228. He quotes at this point Andrzejewski and Lewis 1964.46, who note, as an example: "the Somali classify their poems into various distinct types, each of which has its own specific name. It seems that their classification is mainly based on two prosodic factors: the type of tune to which the poem is chanted or sung, and the rhythmic pattern of the words." The formulation of Ben-Amos may be compared with that of Aristotle Poetics 1447b9-23, as discussed at Ch.1§15.
§38n1. Allen 1973.99-101, who prefers in the end not to use the word rhythm (p. 101). I continue to use it here in the sense of "a system that operates in terms of stress (duration or intensity or both)."
§38n2. N 1974.145. Parts of this statement are already quoted in Allen 1973.14, 258 (cf. also p. 13) with the bibliographical tag "Nagy 1970b," listed as "Monograph (unpub.) on Indo-European metrics" (Allen, p. 378). Without Allen's generous acknowledgment, the existence of an unpublished 1970 version of N 1974 would not be a matter of public record (in N 1979b.629n1 I list those to whom I had sent copies of this unpublished version).
§38n3. On the fundamental role of parallelisms and repetitions in differentiating what I am calling here SONG and SPEECH, see again Guillén 1985.93-121, especially pp. 103-104 with reference to the work of [Zcaron]irmunskij 1965 following Steinitz 1934.
§39n1. There is a particularly interesting example cited by Allen 1973.259n1: in the Luganda traditions of accompaniment, short syllables are regularly accompanied by one drumbeat and long syllables, by two drumbeats. See also Ong 1977b.
§39n2. Allen, pp. 99-101.
§39n3. I cite the formulation of Allen, p. 100 (where "stress" is intended to include the components of duration and intensity: cf. Ch.1§33n1): "Implicitly or explicitly underlying this identification of stress as the basis of rhythm is the conception of rhythm as movement, and of stress, in the production of audible linguistic phenomena, as the motor activity par excellence." Cf. Wylie and Margolin 1981.115, quoting from a 1906 description of Faroese ballad performance and dancing, where the dancers coordinate their voices and movements: "What, moreover, should be well looked after in the ballad singing is to 'get the word under the foot', as the old ones used to say. One gets the word under the foot when one stresses one word or syllable at the same time that one steps along with the foot."
§39n4. In light of this image I would translate skhêma as 'dance-figure' in Plato Laws 654e and 655a, despite the fact, noted by Barker 1984.142n60, that the usage of Plato does not restrict skhêma to the context of the dance. For the notion of skhêmata as 'dance-figures', postures and gestures that represent, see Barker 119n10 on Xenophon Symposium 2.15. With reference to skhêma as a pose that represents, D. Arnold draws my attention to the use of the verb ekhô, from which the noun skhêma is derived, in Herodotus 1.31.5: at the moment that the brothers Kleobis and Biton die after their ordeal in the service of the goddess Hera, it is said that en teleï toutôi eschonto 'they were held fast in this sacred accomplishment' (the context is discussed further at Ch.8§46n2). The narrative goes on to say that this precise moment, with the pose of the brothers in perfect synchrony, is captured as if in a "freeze frame" by the people of Argos, who witnessed the event: they commission the making of eikones 'images' of the brothers, presumably in their completed pose, which they dedicate at Delphi (Herodotus ibid.).
§39n5. This formulation allows for the possibility that some rhythmical types, in the process of becoming purely metrical types, will have developed into a state of incompatibility with dance or instrumental accompaniment or both. On Aristotle Poetics 1448b21-22, where meters are described as moria 'parts' of rhythms, Hardison 1981.71n4 observes: "The passage simply asserts that meters share ('are parts of') certain forms derived from dance music ('the rhythms')." This observation is conditioned by Aristotle's description of a particular meter used for imitating speech, the trochaic tetrameter catalectic, as a rhythm originally associated with dance that later became displaced by another meter, the iambic trimeter, which was dissociated from dance: Poetics 1449a22-24, Rhetoric 1404a31-33. See further at Ch.1§53. Lucas 1978.86 emphasizes the fact that the scholia to Aristophanes Clouds1352 mention dancing to tetrameters. Aristotle's linking of the trochaic tetrameter catalectic with dance may well be extrapolated from such self-references as in Archilochus F 120 W, on which see Ch.1§9. See also Menander Dyskolos 879, with a reference to the accompaniment of iambic tetrameters catalectic by the aulos 'reed'.
§40n1. Cf. Ch.3§39, Ch.3§38n3. For a pioneering discussion of the relationship between pitch accentuation in the ancient Greek language and melodic patterns in ancient Greek song, see Allen 1973.231-234, especially p. 233, where he cites such testimony as that of Aristoxenus Harmonics 1.10 (and following) concerning the difference beween (1) the sunekhês 'continuous' pattern, with gradual shifts of tone in the accentual patterns of everyday speech and (2) the diastêmatikê 'intervallic' pattern, with stylized shifts of tone in song, by way of intervals. According to Aristoxenus (ibid.), the diastêmatikê pattern is singing, not speaking. Aristides Quintilianus, pp. 5.25-6.7 (ed. Winnington-Ingram), posits an intermediate pattern, between the sunekhês and the diastêmatikê, for the recitation of poetry. For similar evidence in Indic melodic traditions and elsewhere concerning the distinctions between tone, that is, pitch accent, and melody, see Allen, pp. 233-234. On zones of overlap and non-overlap in patterns of tone and melody, see Allen, p. 234. On the difference between tone and intonation, see Ch.1§33n1. I expect that languages with fixed patterns in tone would generate melodic traditions different from languages with patterns in intonation only. Cf. Ch.1§40n1.
§40n2. Nettl 1956.136. Cf. also Bright 1963, especially p. 27; also Merriam 1964.285. There are important elaborations in Nettl 1964.281-292; note in particular his analysis of the correlation between the pattern of strong word-initial accent in the Czech language and the pattern of stressed notes beginning musical phrases in Czech folk music, both vocal and instrumental (1964.283); also his observation that, in English folk songs, the melodic contour "tends to descend at the end of a section, phrase, sentence, or song," corresponding to intonational patterns in the language.
§41n1. Sachs 1937.181-203, 1943.30-43.
§41n2. Cf. Herskovits and Herskovits 1947 on Trinidad melodies: "But not all melodies are rephrasings of old ones. Sometimes a tune heard, a European tune, can be 'swung' into a desired rhythm, with perhaps a change of a few measures, or no change at all. In this case, the words to a traditional song might be joined to the new melody, or a proverb might be used and to it added lines from older songs."
§41n3. Note the description of "logogenic" melodies: they are "narrow of range, using small intervals," whereas corresponding dances are "tight, controlled, expressed through narrow steps" (Merriam 1964.253). See Bake 1957.200 on the Indic tradition of the bhâsikasvara 'speech tone', which has the narrowest pitch compass and is employed, according to tradition, in performing the words of the (White) Yajur Veda. Note too the following formulation: "The melodic line follows the text in every detail; the words prescribe the rhythm and the flow; there is one note to each syllable, pitch is independent of duration. One might say that the melody only supports the words" (Bake ibid.; cf. West 1981.115 and 116, who draws particular attention to the old three-pitch and four-pitch patterns).
§41n4. In the case of Balinese music lessons for the young, Merriam 1964.152 notes: "Those instruments which do not play the melody are ignored for the moment, for the melody must be learned first." On patterns of primary convergence and secondary divergence between SONG and SPEECH, cf. the bibliography assembled by Nettl 1964.290-291. Cf. also the discussion of the factor of "tension" in Allen 1973.110-112.
§42n1. Cf. Allen, p. 111: "One could envisage a form of which the pattern is determined by some prosodic feature x, such that there is another feature y whose distribution in the language is partially coincident with that of x. In such a situation one could speak of tension between x and y where the two factors failed to coincide in composition, and of 'concord' or 'harmony' where they coincided and so reinforced the metrical pattern; and such a 'counterpoint' between the patterns of the two features could arguably be manipulated by the poet for artistic ends."
§44n1. Cf. Ch.1§39 and following.
§44n2. Dale 1969.166.
§45n1. Dale 1969.168. For reinforcement of this view on the level of testimony about the actual performance of song, see Pratinas PMG 708 (in Athenaeus 617b-f) and Plato Republic 398d.
§45n2. Cf. also Plato Republic 400a and Cratylus 424c. See the comments on these and other passages by Pöhlmann 1960.30.
§45n3. On patterns of primary convergence and secondary divergence between SONG and SPEECH, I cite again Nettl 1964.281-292.
§46n1. Cf. Koller 1954 on the inherited concept of mîmêsis; for a balanced updating of Koller's synthesis, addressing the criticism of Else 1958, see Nehamas 1982. Following Halliwell 1986.110, I concede that the semantic range of mimesis was shifting, even before Plato, away from the notions of reenactment or impersonation, to accommodate such distinct notions as imitation or the reproduction of appearances. As I shall argue presently, however, such distinct notions are more limited in scope. For Plato's views on mimesis, see Halliwell, pp. 116-122 (also p. 53). As for Aristotle, Halliwell, p. 128, begins his account by mentioning as a possibility "that Aristotle's guiding notion of mimesis is implicitly that of enactment: poetry proper (which may include some works in prose) does not describe, narrate or offer argument, but dramatises and embodies human speech and action."
§47n1. In the case of Lysias 6.51, mîmeisthai refers to the misuse rather than proper use of a priestly costume by Andocides; still, as Halliwell, p. 113, points out, Andocides is "acting out the part of a priest in full."
§47n2. Cf. Halliwell, p. 114, on the nuances of mimesis in Aristophanes Women at the Thesmophoria: "Aristophanes' parody involves, and deliberately confuses, both an ordinary usage of mimesis terms (for impersonation) and a newly developing application of the language of mimesis to the fictional status of dramatic poetry."
§47n3. Further discussion of this passage at Ch.13§12. Cf. the discussion of Royce 1977.73, including this interesting quotation from Boas 1944.14-15 concerning the dance traditions of the Kwakiutl: "In the Cannibal Dance, the women's War Dance, and some others, there is a fixed fundamental gesture like a basso ostinato that is broken at intervals by special gestures of pantomimic character which is descriptive of the text of the song."
§48n1. For the perspective of a social anthropologist on the reenactment of myth in ritual: Leach 1982.5-6.
§48n2. I single out the helpful commentary of Barker 1984.40n4, especially with respect to the reading krembaliastun at Hymn to Apollo 162, which he interprets as "the locally grown rhythmic form, since rhythm is what krembala were used to emphasize." Barker argues that this passage is "advertising the rhythmic and linguistic versatility of the Delian chorus, who might be asked to perform pieces from any of the literary and musical traditions of Ionia and the islands" (ibid.). Cf. Burkert 1987.54: "Contrary to what both others and I myself have written [Burkert 1985.110], I am inclined now to take this [= lines 162-165] as indicating mimetic elements in [the] performance of choral lyrics." Cf. also Bergren 1982.93.
§49n1. Cf. Aeschylus Libation-Bearers 564 and the commentary of Nehamas 1982.56-57. Cf. also Theognis 367-370, as discussed at Ch.12§70.
§49n2. On which see Sifakis 1986, especially p. 218.
§49n3. Tambiah 1985.123.
§50n1. I cite the description of an all-female ritual, as attested in an Ismaili community south of Mashhad in Eastern Iran, which entails the narration of a story as the central event of a ritual meal (Mills 1982). At crucial moments in the retelling of this story (described as a combination of Aarne-Thompson tale type 480, "The Kind and the Unkind Girls," and 510A, "Cinderella"), the girl who is the chief participant, to whom the story is primarily addressed, has ritual food spooned into her bowl by the widow who tells the story, to which the girl answers yes at each of these crucial moments (as recorded in Mills, pp. 185-186).
§52n1. Cf., for example, Dale 1968 on recitative anapaestic meters in drama.
§53n1. There are parallels in medieval traditions: poetic genres where melody is absent are characterized by patterns of prosodic elaboration that seem to serve as compensation for the lost melodic component: see Zumthor 1972.99. On the old French distinction between dit and chant, see Zumthor, p. 406.
§53n2. On the use of the trochaic tetrameter catalectic as a medium of dialogue, that is, imitated speech, see Pickard-Cambridge 1968.158-160. This is not to say that this meter could not be sung, danced, or instrumentally accompanied: see Pickard-Cambridge, pp. 156-158.
§53n3. We may recall the primary nature of the opposition SONG and SPEECH, as discussed by Ben-Amos (quoted at Ch.1§37).
§53n4. West 1982.78, following Pickard-Cambridge, p. 159, who cites, for example, Sophocles Oedipus at Colonus 887-890, Euripides Herakles 855-874.
§53n5. See Pickard-Cambridge 1968.158-160 (also Comotti 1979.21). It may be misleading to some that West 1982.77 uses recitative to translate parakatalogê. To repeat, I view parakatalogê as an intermediate stage between sung and spoken: the parakatalogê is described as a form of delivery that explicitly contrasts with song in "Aristotle" Problems 19.6. See Ch.1§19.
§53n6. Dale 1968.86 and 208.
§53n7. Dale 1968.25-31. Cf. West 1982.98, 128 (especially n125).
§53n8. As in Euripides Andromache 103-116.
§53n9. In line with this argument it is crucial to note that parakatalogê is incompatible with dance, as Rossi 1978 has argued on the basis of Aristophanes Wasps1528-1537. To put it more accurately: parakatalogê is compatible only with a special kind of stylized dance, a mimesis of dance (in this sense, a mimesis of a mimesis), not with dance itself. That is to say, parakatalogê is one stage removed from dance, just as it is one stage removed from singing. Dance becomes reduced, just as melody, in a format of reduced song. Moreover, it may well be that the ritual content itself is correspondingly reduced.
§54n1. Full discussion in Ch.8.
§54n2. We may also adduce the prose of Gorgias (82 DK). For a parallel phenomenon in medieval traditions, where poetic compositions can be subjected to a conscious process of dérimage or "un-rhyming" into prose, see Zumthor 1972.99-100; also Kittay and Godzich 1987. In light of a distinctly juridical function associated with much of early medieval prose, and the fact that the form of an "un-rhymed" composition is perceived as conveying the content of a different level of truth-value from the "rhymed" (Zumthor, p. 98), it is interesting to compare the juridical dimension of early Greek prose authors like Herodotus, as discussed at Ch.9.
§54n3. On which see Ch.11§32. We find a parallel in the medieval genre of the chantefable, such as Aucassin et Nicolette, with alternating song (the melody of the laisses has been preserved) and prose: see Vance 1986.161-163. I note in particular the following summary: "Prose is unmarked speech 'at large', while verse is the marked speech of a social constituency with precise boundaries" (Vance, p. 163). For the Chinese analogue known as chu-kung-tiao, see Chen 1976.
§56n1. For a convenient metrical overview, see the analysis of Snell in the SM edition of Pindaric fragments, pp. 162-174. In only one case, Pindar Olympian13, is there a coexistence of the two types within one composition (Aeolic modulating into dactylo-epitrite; cf. Bacchylides Epinician 3). For an exceptional case of a Pindaric song composed in neither Aeolic nor in dactylo-epitrite meters, I cite Olympian2, composed in Ionic meters.
§57n1. Cf. Appendix Appendix §2n1 and following.
§58n1. On the concept of colon, see Appendix Appendix §1. For stikhos in the sense of verse, cf. Aristophanes Frogs1239. On the concept of a distiction between the monodic and the choral medium, see Ch.3§5.
§58n2. Cf. West 1982.43. It is important, however, not to confuse synartetic with strophic and asynartetic with stichic, in that the category of asynartetic accommodates not only verse but also strophes where the colon is clearly delineated; conversely, the category of synartetic does not accommodate strophes where the colon is clearly delineated. Granted, there are cases where the direction of development is from asynartetic to synartetic (cf. Wilamowitz 1921.421). Still, it hardly follows that the constituents of the strophe are built from the constituents of the verse. As the discussion proceeds, it in fact becomes evident that the direction of development is the reverse: from colon to verse portion, not from verse portion to colon. In any case asynartetic is a category that is roughly half-way between strophic and stichic. Cf. Ch.13§27.
§59n1. Reduction of melody: Ch.1§19; reduction of dance: Ch.1§53n9.
§59n2. Cf. Rossi 1978 on Aristophanes Wasps1528-1537.
§60n1. On the Alexandrian poetic practice of generalizing units taken from the synartetic framework of strophic song into the stichic units of verse in poetry, see Rossi 1971.86. For later developments in the genres of the nomos and the dithurambos, where the principle of strophic responsion is abandoned, see Gentili 1985.35. As it is pointed out in "Aristotle" Problems 19.15, the abandonment of responsion entails greater freedom for experimentation in both the rhythms and the melodies.
§60n2. The meter in question is glyc@2da, on which see Appendix Appendix §3n5. All of Book II of the canonical Sapphic corpus was composed in this meter: Hephaestion 7.7, p. 23.14-17 Consbruch.
§60n3. For example, glyc@da in Sappho F 94 V, on which see Appendix Appendix §3n4.
§60n4. Cf. West 1973.188: "If there was epic or heroic balladry in (say) 1600 [B.C.], its characteristic verse was most likely the glyconic [= glyc], whose cognates are used in Sanskrit and Slavic epic." On the Middle High German evidence for melodic traditions of epic sung in strophic form, see Brunner 1970.160.
§61n1. Cf. Appendix Appendix §21 and following. For the metrical symbols, see Appendix Appendix §2n1.
§61n2. See Appendix Appendix §21 and following.
§62n1. Cf. Rosenmeyer 1968.230. One apparent exception is Pindar Pythian9.4 , 12. Another is Pythian1.92 (_oo_oo_oo_|oo_oo|__o|__, on which see Gentili and Giannini 1977.17), where part of the sequence looks like a dactylic hexameter; it is not in fact a hexameter since there is no correspondence with the final word boundaries of hexameter. There is an analogous situation in Pindar Nemean9: at the beginning of each strophe is a unit _oo_oo___oo_oo__ matching the hexameter in its initial and final word boundaries but clashing with the primary caesura patterns of hexameter.
§63n1. Cf. Klusen 1969.72-83, cited in an interesting discussion by Rösler 1980.104n176. As a counterweight to the notion of gesunkenes Kulturgut in Klusen's work, see Bausinger 1980.41-55.
§63n2. Cf. also Satyrus in Oxyrhynchus Papyri ix no. 1176 fr. 39 col. xix.
§63n3. On which see Ch.1§53.
§63n4. See, for example, Dale 1968.25-31, especially p. 29, on phraseological and prosodic irregularities in the sung varieties of dactylic hexameter--which are irregularities only from the standpoint of the regularities in the nonsung variety. Cf. West 1982.98, 128 (especially n125). See also Zumthor 1972.99 on a comparable situation in medieval European traditions: "Dans les genres non chantés, le perfectionnement des effets sonores semble une sorte de compensation de la perte de la mélodie." As for Zumthor's list (ibid.) of formal features that serve to differentiate poetry from song, I should emphasize that all these formal features are potentially present in song.
§63n5. See N 1979b.614-619.
§64n1. For more on the notion of mode, see Ch.3§1n1.