In earlier work, my starting point has been the central comparative insight of Milman Parry and Albert Lord, gleaned from their fieldwork in South Slavic oral epic traditions, that composition and performance are aspects of the same process in the making of Homeric poetry. [2] Let us continue to call this process composition-in-performance. Starting with the comparative evidence about composition-in-performance, I added the internal Greek evidence about the early diffusion of the Iliad and Odyssey in the archaic period of Greece, positing a model for the development of Homeric poetry that requires not two but three interacting aspects of production: composition, performance, and diffusion. [3] Here I move on to apply comparative evidence for the interaction of composition, performance, and diffusion in attested living oral epic traditions.

My original reasons for concentrating on the role of diffusion-in-performance in the development of Homeric poetry had to do with the need to reconcile the comparative insight of Parry and Lord about composition-in-performance with the historical reality of an integral and unified Homeric text inherited from the ancient world. How the concept of diffusion helps to account for Homeric textuality is a question that will be taken up presently. But first, let us consider the implications of the historical “given,” the survival of the Homeric text.

For some Classicists, the very nature of this written text has been a source of extreme skepticism concerning the validity of applying comparative insights about oral poetics. It is the opinion of not a few of these skeptics that the artistry, cohesiveness, and sheer monumentality of the Iliad and Odyssey rule out the role of oral poetics — and supposedly prove that such marvels of artistic achievement must have required the technology of writing. [4] There are others who even go so far as to argue that Greek alphabetical writing was devised primarily for the purpose of writing down Homeric poetry. [5] Such attitudes toward Homeric poetry tend to be associated with a more general view, shared by some anthropologists, that writing is the basic prerequisite for a major breakthrough in human cognitive capacity, providing the key impetus toward creative thinking and critical judgment. [6] For such skeptics, then, oral poetry and literacy are clearly incompatible with each other.

For other skeptics, oral poetry may not after all be incompatible with literacy, provided we may assume that the art of oral poetry became appropriated altogether by the art of written poetry: this way, we may allow for an oral heritage in the Homeric tradition, but whatever we admire as high art in this tradition must still be attributed to literate authorship. [7]

Either way, whether or not oral poetry is supposed to be compatible with literacy, both these lines of skeptical thinking avoid the comparative insight of Parry and Lord about composition-in-performance, and they assume either explicitly or implicitly that the technology of writing was key to the composition of the actual Homeric text. My own position is that there is no proof that the technology of alphabetic writing was needed for either the composition or the performance of the Homeric poems. [8]

Given that the poetry of the Iliad and Odyssey has indeed survived as a written text, Albert Lord offered a solution for retaining the model of composition-in-performance by postulating that these poems had been dictated. [9] There have been recent attempts to extend this dictation theory, [10] but they run into problems when it comes to explaining the early diffusion of what we now describe as the Iliad and Odyssey in archaic Greece on both sides of the Aegean — a process that some claim was already under way as early as the fourth quarter of the seventh century before our era. [11] Any pattern of diffusion, if indeed it is to be put at so early a date, can hardly be ascribed to a hypothetical proliferation of a plethora of manuscripts, in view of the existing physical limitations on materials available for writing down, let alone circulating, compositions of such monumental size as the Iliad or Odyssey. [12] We must also reckon with the rudimentary status of writing as a technology in that period of Greek history. [13]

One solution is to imagine a situation where a single hypothetical dictated text becomes the prized possession of a special group of performers. [14] Although this modified dictation theory offers some advantages in retaining the factor of performance, there are major problems with it. For one thing, it leaves unexplained a basic question: how exactly was such a dictated text supposed to be used for the process of performance? How could a dictated text automatically become a script, a prompt, for the performer who dictated it, let alone for any other performer? In fact, such a solution does not even mesh with Albert Lord’s actual experience with the phenomenon of dictation in the context of genuine living oral traditions, where the writing down of any given composition-in-performance in effect eliminates the performability of that particular composition. In terms of such a modified dictation theory, moreover, the technology of writing has to be invoked not only for the performance of the Iliad and Odyssey but even for the ultimate composition of these poems, to the extent that the text is imagined to achieve its status as text at the very moment that dictation transforms a composition-in-performance into a script, as it were.

Lord’s original theory of Homeric dictation does not leave room for the use of the dictated text as a mnemonic device for future performances by the singer who dictated it. Lord himself puts it this way: “Someone may suggest that it [= writing] would be a mnemonic device, but this too is unrealistic. The singer has no need of a mnemonic device in a manner of singing that was designed to fill his needs without such written aids.” [15] Following Lord’s reasoning, Raphael Sealey argues that “the singers would hardly feel the slightest obligation to keep to the written text.” [16] Sealey goes on to reject the notion of such a written text, to which he refers as a hypothetical “bardic text”: he argues that, if a composition like the Iliad had been preserved by way of “bardic texts” in the eighth century, then it would have been preserved “by inferior poets.” [17] “But audiences would surely prefer better poets,” he concludes, “and a poem preserved primarily in ‘bardic texts’ would be likely to perish for want of popularity.” [18] Reflecting on the observations made by Albert Lord about actual dictations taken in fieldwork from the South Slavic oral epic traditions, [19] Minna Skafte Jensen argues along similar lines:

I agree with this line of reasoning, at least as far as it applies to the eighth century, the period of Homeric dictation according to the dictation theory as we have seen it formulated so far. As we will find, however, attitudes towards the technology of writing in later periods may well have changed, so that a written version, though not necessarily a dictated version, may indeed in the course of time come to be perceived as “as specially important thing.”

It is not so much that the use of a text as a prompt for performance is unimaginable, once such a text exists. [21] But an even more basic question is, how would such a hypothetical text be conceived in eighth-century Greece, that is, at the earliest stages in the history of this new technology of alphabetic writing? Also, how would we imagine that such a text ever achieved its status as text, starting from the very moment that dictation supposedly transformed a composition-in-performance into a transcript, as it were? I ask these questions because, in terms of some current formulations of a dictation theory, the technology of alphabetic writing has to be invoked not only for the performance of the Iliad and Odyssey but even for the ultimate composition of these poems, to the extent that dictation supposedly creates a basic text. [22]

There are also problems with the very concept of a transcript, a script, or a prompt, supposedly coming into being in the eighth century BCE. Let us consider the earliest attested uses of alphabetic writing in Greece during the first millennium, BCE: starting with the eighth century, we see brief poetic utterances being inscribed in stone. On the basis of these early poetic inscriptions, it can be argued that writing was used not for the actual composition of the utterances being inscribed: “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.” [23] In other words, it can be argued that writing was needed for the recording but not for the actual composition of early poetic inscriptions.

We may ask again: was writing needed for the performance of poetry? As I will now argue, the language of the earliest inscribed utterances makes it clear that writing was being used as an equivalent to performance, not as a means for performance. It is evident from the language of the earliest inscriptions from the eighth century and thereafter, and the pattern holds all the way till 550 BCE or so, that the speech-act of performance was thought to be inherent in the given inscription itself, which normally communicates in the first person, as if it were a talking object. [24] In this earliest attested phase of alphabetic writing, the inscription is not a transcript but a figurative performance, a speech-act that delivers its own message in the first person. [25] It is only after 550 BCE or so that the language of the inscriptions begins to reveal lapses into a mode of talking that is not strictly inherent in the object inscribed, so that the generic inscription now verges on becoming a transcript of an utterance, poetic or otherwise, instead of being the equivalent of the utterance itself. [26] By transcript I mean the writing down of a composition-in-performance not as a performance per se but as a potential aid to performance. And it is only in this later period, after 550 BCE or so, that we begin to see examples of the use of writing for purposes of transcribing any given composition and controlling the circumstances of any given performance. [27]

Let us return to the subject of the earliest phases of Greek poetic inscriptions, starting with the eighth century BCE. For an ancient reader to read such inscriptions out loud is to participate passively in this fait accompli, in that the reader’s voice is being lent to the speech-act which is in this case the very act of writing down the poetic utterance. [28] To repeat, the Greek poetic inscription in the earliest period, before 550 BCE, is not conceived as a transcript of performance, of a short poem: it is rather conceived as a poem, because it is written down, and because this writing down is conceived as an authoritative equivalent to performance. [29] To read the inscription out loud is to become part of the performance that is the writing down: it is to hear the writing itself, not any live performance. The words of written inscriptions can therefore even be quoted in real live performance, as seems to be the case in the passage from the Iliad where Hector is described as imagining the words that implicitly call out from what sounds like an imaginary epigram (Iliad VII 89–90). [30]

If indeed alphabetic writing was perceived as an equivalent to live performance already in the earliest stages of this technology in ancient Greece, and if indeed it continued to be so perceived down to 550 BCE or so, then it is justifiable to doubt the hypothesis that writing had been used, in its earliest phases, as a medium for recording live performance. I would therefore wish to modify slightly the wording of the suggestion that “the alphabet developed specifically or largely in order to record hexameter poetry.” [31] To record epigraphical poetry: yes, maybe, but not necessarily to record epic poetry. Looking at the period before 550 BCE, we may well ask: why would live epic poetry have to be recorded in the first place? The fact that Homeric poetry was meant to be performed live, and that it continued to be performed live through the Classical period and beyond, remains the primary historical given. [32] So we are still left, I maintain, without any internal Greek evidence to prove that the technology of alphabetic writing, as it existed during its earliest phases in the Greek archaic period, was necessary for the performance of the Homeric poems any more than it was necessary for their composition. [33]

It is in this light that I offered, in my earlier work, a different solution to the historical problem of the Homeric text. My solution combined the comparative evidence about composition and performance in attested living oral poetic traditions with the internal evidence of ancient Greek testimony about the diffusion of Homeric poetry in the archaic period of Greece. The comparative evidence from living oral epic traditions, as we are about to see, helps corroborate the internal evidence about the ancient Greek circumstances of diffusion.

Before we proceed with comparing other epic traditions with those of ancient Greece, however, a few words of background are in order about the internal Greek evidence itself. I offer here a minimalist formulation of two basic concepts, “epic” and “Homer.” For Classicists, the idea of “epic” is clear in its application, if not in its definition. Following the usage of authorities like Aristotle, Hellenists can easily distinguish the poetic art-form of epopoiía ‘making of epic’ (as at the beginning of Aristotle, Poetics, 1447a13) from such other poetic art-forms as tragōidías poíēsis ‘making of tragedy’ (ibid.). The application of Homer’s name to the authorship of the Iliad and Odyssey, the prime examples of Greek epic, is also clear. True, the earliest attested references to Hómēros attribute to him not only the Iliad and Odyssey but also the epics of the so-called Cycle, such as the Cypria and the Little Iliad. [34] In fact, the very concept of kúklos, usually translated as ‘circle’ or ‘Cycle’, stems from the ancient pre-Aristotelian tradition of applying the metaphor of cycle to the sum total of epic poetry, as if all of it were composed by Homer. [35] By the time of Aristotle, however, the epics of the Cycle are conventionally assigned to distinct authors (Poetics 23.1459b1–7). [36] Such eventual disruption in the semantics of the very concept of Cycle is not a matter of common knowledge among contemporary experts in Homer.

What made decisive the differentiating of the Iliad and Odyssey from all other epic poems was the influence exerted by the scholars at the Library of Alexandria, particularly by Zenodotus of Ephesus: “it was of the utmost importance for the whole future that [Zenodotus] … accepted the differentiation between these two poems as Homeric and the rest of epic narrative poetry as non-Homeric.” [37] Though there were attempts to narrow down the Homeric corpus even further, as when scholars known as the “separators” or khōrízontes tried to separate the authorship of the Odyssey from that of the Iliad (Proclus p. 102.2–3 Allen), the Alexandrian verdict on Homer as the author of the Iliad and Odyssey held firm in the ancient world. The “Homeric Question,” as reformulated in the Renaissance and thereafter, must be viewed against this background; so also the comparative insights pioneered by Parry and Lord.

The progressive restriction of what exactly in Greek epic is to be attributed to Homer can be connected with the historical process that I have just highlighted, to wit, the relatively early diffusion of the Iliad and Odyssey throughout the Greek-speaking world. In my earlier work, I adduced archaeological evidence, as assembled by Anthony Snodgrass, pointing towards a trend of pan-Hellenism that becomes especially pronounced in archaic Greece in the eighth century before our era and thereafter. [38] The epic tradition of Homer, as Snodgrass inferred from the early proliferation of the Iliad and Odyssey, was a reflex of this trend of pan-Hellenism. [39] I extended Snodgrass’s concept of pan-Hellenism, setting it up “as a hermeneutic model to help explain the nature of Homeric poetry, in that one can envisage as aspects of a single process the ongoing recomposition and diffusion of the Iliad and Odyssey.” [40] I had called this model for the text-fixation of Homeric tradition “evolutionary,” without intending any Darwinian implications about progressive superiority. [41] According to this evolutionary model, as I have formulated it in my earlier work, the process of composition-in-performance, which is a matter of recomposition in each performance, can be expected to be directly affected by the degree of diffusion, that is, the extent to which a given tradition of composition has a chance to be performed in a varying spectrum of narrower or broader social frameworks. [42] The wider the diffusion, I argued, the fewer opportunities for recomposition, so that the widest possible reception entails, teleologically, the strictest possible degree of adherence to a normative and unified version. [43]

I continue to describe as text-fixation or textualization the process whereby each composition-in-performance becomes progressively less changeable in the course of diffusion — with the proviso that we understand text here in a metaphorical sense. [44] The fixity of such a “text,” of course, does not necessarily mean that the process of composition-in-performance — let us continue to call it recomposition — has been stopped altogether. So long as the oral tradition is alive, some degree of ongoing recomposition is still possible in each performance, even if the tradition itself proclaims its own absolute fixity. A case in point is the so-called “Invocation of the Bagre,” a “hymn” sung among the LoDagaa of Northern Ghana. [45] It is clear that the expectation of both the audience and the reciters of the Bagre is that each performance be exactly like every other performance, but empirical observation shows that it is not. Reaching a size of up to 12,000 lines, the Bagre in fact exists in a variety of versions, and the differences among the versions can be considerable. [46] In sum, the rate of retardation or acceleration of change in the process of composition-in-performance depends on the stage of evolution in which we happen to find any given living oral tradition. [47]

In arguing for the notion of a single pan-Hellenic tradition of epic — let us call it Homer — as opposed to a plethora of local traditions, I stressed the relativity of the term pan-Hellenic from an empirical point of view:

In other words, I am arguing that the concept of pan-Hellenism is not at all incompatible with the factor of change. I therefore disagree with the implications of the following assessment:

To repeat, the model of pan-Hellenism is by definition not rigid, not even for Homer.

In terms of such a model, which we may continue to describe as evolutionary, I posit at least five distinct consecutive periods of Homeric transmission, “Five Ages of Homer,” as it were, with each period showing progressively less fluidity and more rigidity: [51]

1. (1) a relatively most fluid period, with no written texts, extending from the early second millennium into the middle of the eighth century in the first millennium. [52]
2. (2) a more formative or “pan-Hellenic” period, still with no written texts, from the middle of the eighth century to the middle of the sixth. [53]
3. (3) a definitive period, centralized in Athens, with potential texts in the sense of transcripts, at any or several points from the middle of the sixth century to the later part of the fourth. [54]
4. (4) a standardizing period, from the later part of the fourth century to the middle of the second; this period starts with the reform of Homeric performance traditions in Athens during the régime of Demetrius of Phaleron, which lasted from 317 to 307 BCE. [55]
5. (5) a relatively most rigid period, from the middle of the second century onward; this period starts with the completion of Aristarchus’ editorial work on the Homeric texts, not long after 150 BCE or so. [56]

A context for the definitive period in my evolutionary model is a pan-Hellenic festival like the Panathenaia at Athens, which served as the formal setting, established by law, of seasonally recurring performances of the Homeric Iliad and Odyssey (cf. Lycurgus Against Leokrates 102). [57] In the next chapter, we will consider in more detail the question of identifying who should be given credit for shaping or, better, reshaping not only this institution of the Panathenaia but also, in particular, the institution of Homeric performances that became a featured event of this festival. At this point it is enough to highlight the involvement of the Peisistratidai — that is, of Peisistratos and his sons — who traced themselves back to the heroic-age Peisistratos, son of Nestor (as portrayed in Odyssey iii) and who ruled Athens as tyrants during the second half of the sixth century BCE. [58] More important for now, I argue in what follows that the Feast of the Panathenaia is a salient example of a distinct pattern of diffusion in oral traditions.

I hope to show from the comparative evidence of various oral epic traditions that there is more than one way to visualize the actual process of diffusion. Besides the pattern of an ever-widening radius of proliferation, with no clearly defined center of diffusion, there is also a more specialized pattern that can be predicated on a functional center point, a centralized context for both the coming together of diverse audiences and the spreading outward of more unified traditions. In other words, a fixed center of diffusion can bring into play both centripetal and centrifugal forces. Such a center point, as we will see, is the seasonally recurring festival of the Panathenaia at Athens.

In general as also in details, an evolutionary model for the text-fixation or textualization of the Homeric tradition is corroborated by the comparative evidence of living oral epic traditions. We may note in particular the results of recent fieldwork in the oral epic traditions of the Indian subcontinent. [59] In what follows, I quote extensively from descriptions of the Indian evidence that were formulated by observers who were not at all concerned with the ancient Greek evidence to which I am now applying them. [60] The degrees of similarity between the empirical observations of the Indian evidence, as I cite them verbatim, and some of my constructs derived from the Greek evidence are to my mind so striking that a casual reading could leave the mistaken impression that the wording of the Indologists was influenced by these constructs.

For example, researchers like Stuart Blackburn who specialize in the oral folk epic traditions of latter-day India have developed, independently of their counterparts who specialize in ancient Greece, the descriptive term pan-Indian, which they correlate with observable patterns of what they call geographical diffusion in epic traditions. [61] Matching the Greek model, the descriptive term “pan-” is used in a relative sense, as we can see from the following explicit restatement of Blackburn’s position in the introduction to the book that he co-edited: “Tracing a narrative pattern that moves from a local hero toward a wider, more pan-Indian identity for the hero/god, [Blackburn] concludes that this change is a response to the differing social groups and contexts encountered as an epic spreads geographically.” [62] In effect, then, Blackburn is positing an ongoing process of recomposition in the making of Indian epic that is analogous to an evolutionary model for the making of ancient Greek epic.

An evolutionary model is applicable also to the two great canonical Sanskrit epics of the Indian subcontinent. [63] These are the Mahābhārata, an epic of truly monumental dimensions which, in its ultimate form, is roughly eight times the size of the Homeric Iliad and Odyssey combined, and the relatively smaller Rāmāyaṇa. The performance traditions that culminated in these two Sanskrit “primary epics” extended well into the second half of the first millennium of our era. [64] It took “several centuries” for both “to reach their final forms,” [65] with the “formative period” of the Mahābhārata estimated at around 400 BCE to 400 CE and of the Rāmāyaṇa, at around 200 BCE to 200 CE. [66]

The relative lateness of text-fixation in the case of these two canonical Sanskrit epics is illustrated by the fact that even the earlier of the two, the Mahābhārata, “began to take shape” at a time when the Vedas, the “priestly literature” of the Brahmin caste which had already formalized the technology of writing, had reached an advanced stage of development. [67] “Viewed against this background the Mahābhārata represents, as it were, a return to the beginning. It was an oral composition; it was purely heroic in character; and it dealt with people and events of which the earlier [Brahmin] literature had taken practically no notice.” [68]

This dichotomy between Vedic and epic can be explained to some degree in terms of caste distinctions: “as the Vedas and their supporting literature were the ‘property’ of the Brahmins, so the epic was the ‘property’ of the Kṣatriya-s, the caste of warriors and princes.” [69] Although the Kṣatriya-s “owned” the epic, it was not composed/performed by them but by a specialized caste called the Sūta-s, performing to the accompaniment of an early form of a string instrument known as the vīṇā. [70] The relationship of these Sūta-s and the Kṣatriya-s is analogous to that of the medieval Cāraṇ court poets and their Rājpūt patrons. [71]

The Kṣatriya background of the Sanskrit epic tradition becomes accretively displaced and covered over, in the course of time, by a Brahmin superstructure, and the very process of this displacement can be interpreted as a sign of fluidity in an oral epic tradition. The pattern of displacement is so pervasive that the very authorship of the Mahābhārata is traditionally attributed to one Kṛṣṇa Dvaipāyana, a Brahmin seer who is also a major character in the plot of the Mahābhārata. [72] Similarly with the Rāmāyaṇa, where the Kṣatriya hero Rāma becomes accretively identified with the Brahmin god Viṣṇu.

Granted, experts have not yet determined to what extent such accretive patterns of displacement took place on the level of oral transmission in the two primary Sanskrit epics. [73] But, more importantly, the fact is that the living oral epic traditions in India today, affording as they do a wealth of evidence for various patterns of diffusion, provide comparative evidence for an evolutionary model that helps explain the actual process of accretion. [74]

As we proceed to consider in some detail the evidence of living oral epic traditions in contemporary India, it is important to stress the explicit role of religion in the very function of epic. For purposes of this presentation, a minimalist working definition of religion will suffice: let us consider it for the moment as simply the interaction of myth and ritual. [75] I propose further to specify “religion” in terms of cult, which I define for the moment as a set of practices combining elements of ritual as well as myth. [76] As one of the most powerful illustrations of the role of religion in the performance of Indian epic, I point to those situations where the performer presupposes the presence of an audience of gods, “watching out for errors in performance.” [77]

For the first specific example of the sort of empirical evidence that has been collected by contemporary researchers concerning the role of cult in the living traditions of epic in India, let us consider the wording of a scholar specializing in Rajasthani epic, who offers the following formulation for the role of epic in the Rajasthani cultural ethos: “concern for propitiating the powerful spirits of those who died untimely deaths continually feeds the epic traditions of the area.” [78] In quoting this description of heroes in the Rajasthani epic traditions, I highlight the word untimely because of its relevance for comparison with the concept of the hero in ancient Greek epic traditions. In the case of the Herakles myth, as adduced by Greek epic itself in the retelling of Iliad XIX 95–133, the theme of Herakles’ unseemliness goes to the very core of the hero’s essence and extends to the essential unseemliness of the main hero of the Iliad, Achilles, who in the end describes himself as the ‘unlikeliest of them all’, pan-a-ṓrios (XXIV 540). [79] This theme of untimeliness in ancient Greek traditions is not restricted to epic: it extends to the concept of heroes in the specific context of their being worshiped in cult, as I have argued extensively elsewhere. [80] As I have also argued, the cult of heroes is a subtext, as it were, for the development of epic traditions about heroes in ancient Greece. [81] Moreover, the relationship between the cult of heroes on a local level and the epic of heroes on a pan-Hellenic level is crucial for coming to terms with the factor of diffusion in the Homeric tradition. [82] As we will see, there are striking analogies in the living traditions of India.

The ancient Greek hero’s untimeliness in myth is to be contrasted with his/her timeliness in ritual, in that the cult of heroes is predicated on the central fact of seasonal recurrence, controlled by the goddess of timeliness herself, Hera. [83] The cults of heroes/ancestors in India too seem to operate on a cyclical principle. Although “the yearly cycle of folk epic performances has yet to be conclusively outlined for different regions and different groups within these regions,” [84] there are isolated cases where we do have specific information: a notable example is the festival of Caitrī, where the performances of epic seem to be connected with the remembrance of ancestors. [85]

Pursuing the question of the relationship between epic traditions and the cult of ancestors or heroes, let us consider in some detail the evidence from Rajasthan, starting with the following summary: “The two major Rajasthani historical epics, Pābūjī and Devnārāyaṇ, appear to have developed out of a tradition of honoring powerful spirits of the dead. A continuing concern about powerful spirits provides the framework in which these epics retain their meaning and vitality.” [86] In the case of the Pābūjī regional epic tradition, it has been argued that it developed out of a local bhomiyā cult. [87] The bhomiyā have been described as “local heroes who died while defending against cattle raids, were commemorated with a carved pillar (showing a rider on a horse), and eventually became gods at the center of a cult.” [88] Pābūjī himself is generally worshiped as a god. [89]

As we will see, the semantic shift from hero to god is a peculiarly Indian phenomenon, characteristic of those stages of epic tradition that have undergone the broadest patterns of diffusion and have thus attained the most normative possible levels of the “Hindu” world view. For purposes of comparison with the ancient Greek evidence, where the distinction between hero and god is for the most part clearly maintained, it is preferable to start with Indian patterns of cult on the most local levels, where we find the most straightforward evidence for the cult of heroes as distinct from gods. Before we begin, however, it is important to note that the ancient Greek distinction between ancestor and hero, unlike the distinction between hero and god, becomes increasingly blurred as we move further and further back in time. [90]

With this caution in place, let us consider further the folk culture of Rajasthan, which has been singled out for its striking differences from the upper-caste Hindu “norm.” [91] I propose to pursue our consideration of the Rajasthani bhomiyā, who are literally the spirits of dead warriors. [92] At the shrine of the bhomiyā,

As for the actual performance of the epic, “the central belief is that singing the hero’s story summons him as a god, whose power is then present to protect the community.” [94] What gives the hero his ultimate power is the actual fact of his death: “the death event operates as the ‘generative point’ for stories in local traditions. It leads to deification, to worship, to a cult, and eventually to a narrative which is ritually performed to invoke the spirits of the dead.” [95]

It can be said in general about the epic traditions of India that their function is a matter of explicit ritual as well as myth: “epic performances ritually protect and cure, while epic narratives express local ideologies and form pathways between regional and pan-Indian mythologies.” [96] For the moment I simply note, for the second time now, the use of the term pan-Indian in this description — a subject to which we will return presently. I also note a point that may not be obvious to those unfamiliar with the perspectives of social anthropology, which is that the element of ritual in such descriptions should not be understood so narrowly as to exclude what we may ordinarily think of as entertainment: “Indian oral epics tend to have performance contexts that are either ritualistic or entertainment-oriented. These two performance contexts exist on a continuum because ritual and entertainment are not mutually exclusive.” [97]

The relation of the local epic to the community is all-important in the Indian traditions: “oral epics in India have that special ability to tell a community’s own story and thus help to create and maintain that community’s self-identity.” [98] Once the local story extends beyond the community, however, there is change in content as well as form. Let us consider the following description of what happens to the theme of the death and deification of local heroes in the context of diffusion: [99]

For the third time now, we note the use of the word pan-Indian in describing the ultimate stages of epic diffusion in India. We may note as well that the application of this term is reserved for the regional level of diffusion and beyond. The categories of regional and subregional are part of an overall taxonomy developed by Stuart Blackburn for the purpose of classifying the relative ranges of diffusion for fifteen samples of living epic traditions in India. The ranges of diffusion for these fifteen selected epic traditions, preceded by categories of description for these ranges, are as follows:

1. 1) local = 10-100 mile range
2. 2) subregional = 100-200 mile range
3. 3) regional = 200-300 mile range
4. 4) supraregional = 400+ mile range. [101]

We may note in particular the observation that the breakthrough of an epic from local to subregional status is promoted by a cult where a large festival is held annually at a single temple. [102]

An ancient Greek analogy that immediately comes to mind is a pan-Hellenic festival like the Panathenaia at Athens, which, as we have already noted, served as the formal setting, established by law, of seasonally recurring performances of the Homeric Iliad and Odyssey (Lycurgus Against Leokrates 102). As the comparative evidence of oral epic traditions in contemporary India shows, the institution of Homeric performances at the Panathenaia can be visualized as a process of diffusion. In other words, diffusion is not restricted to the pattern of an ever-widening radius of proliferation, with no clearly defined center of diffusion. As the Indic comparative evidence shows, there is also a more specialized pattern that can be predicated on a functional center point, bringing into play both centripetal and centrifugal forces. Such a center point, to repeat, can take the form of a centralized context for both the coming together of diverse audiences and the spreading outward of more unified traditions.

For purposes of further comparison, we may observe that the more pan-Indian the epic, the more divergent it seems from the ancient Greek evidence. The actual phenomena of pan-Indianism and pan-Hellenism are comparable, but not so much the results of the respective phenomena of synthesis and diffusion. The most divergent point of comparison is the tendency of the hero’s being elevated to the status of divinity in pan-Indian traditions. I have in mind such phenomena in the Indian evidence as the appropriation of epics having a ritual context by such pan-Indian trends as Vaiṣṇava veneration. [103] True, even where the hero is divinized, there can survive traces of hero-god distinctions: in the Tulu traditions, for example, there is a distinction made between the bhūta or deified dead and the dēvarụ, that is, deities of divine origins. [104] Clearly, the deified dead represent the new and augmented phase of the hero, just one step removed from the status of deities proper. Just as clearly, there are attestations of the next logical step: “new social groups accept the hero as a god and not simply as the deified dead because they have no close link to him by blood or locale.” [105]

All this is not to say that the hero’s elevation to the status of divinity cannot happen on the most local level. [106] Conversely, in the case of the Lorik-Candā epic, which fits Blackburn’s category of “supraregional” epic, [107] its hero and heroine, Lorik and Candā, “are not deified and thus [this epic’s] spread is not associated with that of a religious cult.” [108] There are even more important exceptions: “the Mahābhārata heroes, like the heroes of the Ālhā, do not die in battle, are not deified, and are not widely worshiped. They, too, lack both the conditions and the need for deification.” [109] Still, the general trend of pan-Indian oral epic traditions is the highlighting of the hero’s immortalization and the shading over of his mortality: “as these stories diffuse (even to a limited extent within the local range), they change [highlighting mine]. The hero’s death remains the central narrative event, evoking emotional responses in listeners and explaining the hero’s new status as a god, but it becomes less local history and more narrative convention.” [110] There is even a tendency in Indian traditions to avoid describing the actual death of the hero. [111] In the pan-Hellenic traditions of the Iliad and Odyssey, conversely, the topic of a hero’s immortalization tends to be shaded over, while his mortality and the circumstances of his death are highlighted as the centerpiece of Homeric humanism. [112]

Either way, whatever the direction of shifts in emphasis may be, both the Greek and the Indian traditions seem to become progressively less occasional or ad hoc in the process of diffusion. To discover the occasional or ad hoc applications of ancient Greek epic, of course, is largely a matter of reconstruction or at least of inference from the surviving texts. In the case of Indian epic, on the other hand, there is a great deal of direct evidence about occasionality in the living traditions, and such testimony as we will see affords valuable comparative insights that help better understand the available testimony of the Greek traditions.

Let us begin with Indian evidence about the circumstances of performance and of performer-audience interaction. There are two basic performance-types: song-recitation and dance-drama; [113] dance-drama has been described as “a secondary form in that it only exists where song-recitation also exists.” [114] Performers of epic — singers, musical accompanists, dancers, and ritual specialists — are predominantly from the middle- and low-level castes; by contrast, Classical performance traditions of the Sanskrit epics were “controlled by high-level castes, often Brahmins.” [115] The possibility of a performer’s traveling through different districts seems to be linked with the degree of his professionalism. [116] This phenomenon of professionalization, which seems to be a key to the factor of diffusion, is analogous to the status of the ancient Greek aoidós ‘singer’ as a dēmiourgós, that is, an itinerant artisan (Odyssey xvii 381–385). [117]

In the Indian traditions, the notion of audience is actually more appropriate in the case of professional singers’ performances, whereas some more neutral term like group suits the sort of situation where “non-professional general caste groups sing for the group itself.” [118] For purposes of comparison, the ongoing distinction between audience and group in these descriptions of Indian traditions is pertinent to the scenes of person-to-person or person-to-group interaction in Homeric narrative that seem to mirror the conventions of performer-audience interaction in the “real world” that frames the performance of the narrative. [119] It is also pertinent to the issues raised by Wolfgang Rösler’s Dichter und Gruppe, a work that investigates the reception of archaic Greek lyric in the specific social context of archaic Lesbos. [120] We may ask, for example, on the basis of the comparative evidence, whether the interaction of Alcaeus with his group on one level simply mirrors the performance of the Alcaeus-persona to the audience on another level. [121]

In the case of the general caste group’s “non-professional” performing of epic in the Indian traditions, even this broad category for the aspect of performers has its own structure. There has to be a leader, who generally has had more background in performance than the others, including the mastery of a musical instrument. [122] Potential leaders, who are specialists in a sense, have to compete with one another, and I infer that increasing specialization in the performance of epic is a functional correlate of increasing formalized competition among performers. [123] A comparable correlation of professionalization and competition is discernible in archaic Greek songmaking traditions. [124]

In the course of this brief survey of occasionality in the living epic traditions of India, we may note in passing that epic, as a form of public activity, is performed almost exclusively by male singers. [125] The rarely found exceptions, however, are particularly revealing. For background to the case about to be cited, we may note that the Ahir caste of Uttar Pradesh appropriates an epic known as the Lorik-Candā; [126] this epic “helps to maintain the Ahirs’ image of themselves as a warrior caste.” [127] “It is primarily Ahirs who sponsor performances at occasions such as weddings and the birth of a child. The Lorik-Candā epic is also sung at various festivals, during the harvest season, and at village or town fairs.” [128] In Chhattisgarh, the corresponding epic is called Candainī, and it is with the background of reference to this tradition that we turn to an exceptional case of performance by women. The researcher reports as follows: “One night as I was recording an elderly Gond (tribal) woman singing a variety of narrative songs, she began singing about the wedding of the epic heroine and her first husband. But the woman did not consider this to be Candainī singing.” [129] The narrative content in fact corresponds to Candainī, but the form is different: a distinct rāg ‘tune’ and style. [130] In this case, we find a striking ancient Greek parallel in Sappho fragment 44, the so-called “Wedding of Hector and Andromache”: this song, composed in a meter that is cognate with but distinct from the epic dactylic hexameter, deals in a non-epic manner with themes that are otherwise characteristic of epic. [131] We have here a particularly striking example of the effects of a given occasion on the very nature of epic composition. Just as the song of Sappho about the Wedding of Hector and Andromache is exceptional in the history of Greek literature, so also the song of the elderly Gond woman proved to be exceptional in one particular researcher’s survey of living Indian oral epic traditions. It may well be worth asking whether this discovery about women’s traditions in India would have been possible if the researcher in this case, Joyce Flueckiger, did not happen to be a woman. The question is whether a woman researcher would be deemed by her women informants to be more suitable for the reception of distinctly women’s traditions. [132]

In the many epic traditions of India, there are striking examples of selectivity in choosing not only which topics to highlight or shade over in a given sequence but also which variant of a given topic to use within that sequence. Such choices are tuned to the narrowness or breadth of audience reception. Let us consider two situations, one where the local aspects of an epic tradition have to be highlighted and another where the same aspects are shaded over. We begin with the Kordabbu epic tradition of the Tulu-speaking area of Karnataka, a tradition where parts of the narrative are recited by the possessed priest “in a voice characteristic of the spirits”; this stretch of narrative is marked by a switch from the third to the first person, and is known as the Words of the Hero. [133] “In his performance the possessed priest must not only recite Kordabbu’s story, but also assume his character and dramatically portray his exploits for several hours on end.” [134] This description applies to the Mundala caste. But there is also another performance tradition, called the kōla, maintained by the Nalke caste, which is “a professional bardic caste.” [135] It has been reported about these performers:

I save the most important detail for last: the Nalke “are not likely to elaborate specific details that might offend the sensibilities of a particular group in a village and give rise to a dispute. The Nalke leave the details of the hero’s life and his relationship to other castes to the villagers concerned.” [137] An analogy that immediately comes to mind is the screening out of local traditions from the repertoire of aoidoí ‘singers’ as itinerant artisans in archaic Greece, with the result that the subject matter controlled by such performers becomes a sort of least common denominator appropriate to the most generalized kinds of audience. [138]

In contrast to the distinct non-occasionality of attested ancient Greek epic, we have by now seen a great deal of comparative evidence for occasionality on the level of local performance in the living oral epic traditions of India. There is ample evidence also from the epic traditions of Central Asia. [139] It will suffice here to quote a particularly revealing description of occasionality in Kirghiz epic traditions from a report published in 1885 by a pioneer in the study of oral epic traditions, Wilhelm Radloff:

Such testimony is pertinent to the comparative information about the living oral epic traditions in Africa, where we see a similar correlation of occasionality with local contexts. Let us consider the epic traditions of Manding society, marked by “both a vigorous pan-cultural tradition and a constant pull toward diversity,” crossing as it does several linguistic and modern political boundaries. [142] “The Manding peoples believe that their oral stories retell the experiences of their common past, yet the diversity of their multiforms shows the ability of these stories to adapt to changes of time and locality.” [143] The centerpiece of Manding oral poetry is an epic tradition about a historical figure called Sunjata, a powerful chieftain whose lifetime is historically dated to the thirteenth century CE and who is recognized as the founder of the Manding empire. [144] Recorded versions of the Sunjata narrative range in length from a single evening’s performance to a thirty-hour stretch. [145] In some of these recorded versions, we can find explicit documentation of the singer’s selective use of available narrative versions that tie in directly with such details as the genealogies of members of his audience. [146] The degree of occasionality in the performance of Sunjata epic traditions justifies a formulation such as this one: “The epic is more than the tale of its characters; it is at the same time about its audience.” [147]

To the existing comparative information about epic in Africa we may add still further information, from the realm of praise poetry. When we take an overall look at the evidence collected in Africa, it appears that praise poetry, in the process of diffusion from local towards more regional contexts of performance, progressively takes on the characteristics of what we might otherwise call epic. This trend is markedly noticeable, for example, in the traditions of praise poetry in Xhosa society. [148] Here we may adduce the internal evidence of Greek civilization concerning the relationship of epic with praise poetry. Following the formulation of Aristotle’s Poetics (1448b27, 32–34), who derives epic from praise poetry, I have argued elsewhere that the form and content of a Greek poetic tradition that calls itself aînos or ‘praise’, as represented by the victory odes of Pindar, can be reconstructed as a basis for the development of what we know as epic. [149] In line with my intent to avoid monogenetic theories for the origins of Greek epic, [150] it is important to stress that praise poetry can be reconstructed as a basis, not the basis, for the development of epic. [151] Still, the internal testimony of ancient Greek epic itself implies the outright derivation of epic from praise. We may note references made by Greek epic to primal scenes of praise and blame poetry, as we see in the brief retelling of the Judgment of Paris scene in Iliad XXIV 29–30, where the Homeric tradition itself represents the genesis of epic in terms of a primal opposition of praise poetry to blame poetry. [152]

As we look more closely at the comparative evidence concerning the relationship of praise poetry and epic, we can find further justification for deriving Greek epic, at least in part, from praise poetry. In my earlier work, I studied in great detail the occasional nature of the ancient Greek praise-poem or aînos. [153] Here I simply compare this feature of occasionality in ancient Greek praise-poetry with the occasionality of epic in the oral traditions of India, especially on the more local levels. In India, we find clear instances where plot variation is radically conditioned by the nature of the audience. [154] Such conditioning reveals the dependence of the performers on their audiences. Let us take as an example of such performers the Nayaks, a caste of hereditary singers of the Pābūjī epic tradition found primarily in central and south Rajasthan, who “circulate from village to village on a yearly beat seeking patrons.” [155] The musical instruments of the professional performers tend to be chordophonic, requiring rigorous training. [156] In the commissioning of a Nayak performance of the Pābūjī epic, “the patron’s devotion is the most important measure of the performance.” [157] The patronage can occur on the level of festivals, but most often on the level of the village; “patrons may sponsor a performance for one night or a series of nights.” [158] One motivation for a sponsor’s undertaking of a sponsorship is to fulfill a vow. [159] Such a relationship between patron and poet offers a wide spectrum of comparative insights into the sociology, as it were, of praise poetry in ancient Greece.

The ad hoc orientation of the ancient Greek praise-poem or aînos, with its persistent internal references to the occasion of its performance and to the expectations of its audience, stands in marked contrast to the stance taken in the Homeric tradition of epic, which programmatically shades over any reference to any specific occasion of performance and thus implies that it is worthy of universal acceptance, that is, of unconditional reception. [160] It is as if the epic of Homer had outgrown the need for occasionality of performance. Similarly in the praise poetry of the Xhosa, the phenomenon of diffusion entails the widening of perspective in the content of praise:

The wording here, with emphasis on text as a metaphor for composition in oral poetics, is apt, in the sense that the authoritativeness of such a composition is made analogous to the potential authoritativeness of a written text. And so we come back full circle to our point of departure, which is the historical reality of the Homeric text. We have yet to consider the text as text, but by now we can see, at least in its broad outlines, the process of evolution that led to this reality.