Introduction

Just as the Greek language is cognate with the Sanskrit language, so also the meters of Greek Lyric are cognate with the meters of Sanskrit Vedic. This hypothesis was first propounded by Meillet, [1] who applied the linguistic techniques of comparative reconstruction [2] to Greek and Indic versification. If it is true that the native meters of these Indo- European languages are related, it may be possible to find correspondences of traditional phrases embedded in these meters. My point of departure is the phraseological correspondence, already noticed by Kuhn in 1853, [3] between Sapphic/Homeric κλέοc ἄφθιτον and Rig-Vedic śráva(s) ákṣitam. Both the Greek and the Indic expression are conventionally interpreted as ‘imperishable fame’, and both can be reconstructed to an identical prototype, *klewos n̥dhg^whitom. [4] I propose to show that the metrical contexts of these cognate phrases are also cognate. My purpose is to correlate Meillet’s theory of Indo-European meter with the reconstruction of an Indo-European poetic expression. {1|2} This monograph, then, is a set of metrical studies threaded together by *klewos n̥dhg^whitom.

If I should succeed in showing that a metrical correspondence can be matched with a phraseological correspondence in the traditional poetry of the Hellenic and Indic civilizations, Meillet’s theory of a common Indo-European metrical heritage would be considerably strengthened. To my mind, admittedly, the comparative evidence of meters native to other Indo-European languages has already vindicated Meillet. The Indo-European origin of Slavic meter has been demonstrated by Jakobson [5] and confirmed by Watkins. [6] In addition to Indic, Greek, and Slavic, Watkins has argued convincingly that Celtic meter is also related. [7] More recently, Cole has adduced the evidence of the Saturnian to argue that Italic meter is a fifth cognate. [8] What I seek now is simply further confirmation in another dimension—one which goes beyond meter and which I am prepared to call phraseology, for want of a more specific term that would not mislead. Since Meillet was working with Greek and Indic meter only, and since the phrase *klewos n̥dhg^whitom happens to survive in Greek and Indic diction only, my work ends up being a reappraisal of Meillet’s arguments on his own terms.

As things now stand, Meillet’s theory of a {2|3} common heritage for Greek and Indic meter is weaker on the Greek side—at least it is weak enough to be ignored by most students of Greek meter. [9] Among the standard works, only the masterly Griechische Verskunst of Wilamowitz is openly sympathetic to the idea of Indo-European meter. [10] Wilamowitz was writing before Meillet, however, and his thinking is based more on intuition than facts. I lay such emphasis on Greek metricians not because Indic metricians have generally taken up Meillet’s theory (they have not), but because I myself think that much more needs to be said about the evolution of Greek meter.

The inherent difficulties of Greek meter emerge more clearly as we contrast it with Indic meter. Unlike the Indic Rig-Veda, which features about half a dozen basic types of verse and a dozen basic types of stanza, the attested compositions in the tradition of Greek Lyric are notorious for their dazzling variety in verses and stanzas of every imaginable size and shape. Then there is the problem of attestation. The Rig-Veda consists of over a thousand religious hymns, some of them composed as far back in time as the second millennium B.C. and all of them preserved and transmitted with the greatest care in the ritual system of a tight-knit priestly society. By contrast, most of Greek Lyric comes {3|4} down to us in desperately small bits and pieces. Moreover, of what little we have there is nothing anterior to the era of poets like Archilochos. In a word, we have practically no concrete idea of what Greek Lyric might have been like before the early seventh century B.C. Serious canonization of the texts starts much later, in the period of Alexandrian scholarship, [11] and by that time it was already too late for many of the less famous poems to survive. Even worse, the Alexandrian canon has been irretrievably lost, for the most part.

As I will take pains to show in the succeeding chapters, the meters of the Rig-Veda are few in number but flexible in pattern, while those of Greek Lyric are many but rigid. Of course, trends are by nature more easily discernible in flexible patterns. Statistical analysis is possible, leading to all sorts of valuable conclusions about the evolution of this meter or that. Such analysis has in fact been a hallmark of Rig-Vedic metrical investigation, and the results have been impressive. We may develop a fairly good idea of proto-Indic meter simply from the internal evidence of Rig-Vedic meter as it is described in the meticulous studies of Oldenberg and Arnold. [12] On the other hand, trends may often only be intuited in the rigid meters of Greek Lyric. For example, if two {4|5} rigid meters X and Y are identical except for one distinctive feature, where X has a plus and Y has a minus, it seems reasonable to derive them from a single, more flexible, meter where the plus/minus distinction does not exist. But actual proof for such a prototype would have to come from comparative evidence. “A history of Greek Metric,” as A. M. Dale wryly remarks, “is limited to recording the appearance of a great variety of meters, each sprung perfect like Athena in full panoply from the head of Zeus, and to tracing modifications of these in the practice of later poets.” [13] In short, it is inherently difficult to reconstruct the prehistory of Greek Lyric meters on the sole basis of the attested internal evidence.

Meillet was faced with still another great difficulty in his comparative study of Greek meters. Unlike Indic Epic, which dates from a much later era than the Rig-Veda and the meters of which are easily derivable from those already found in the Rig-Veda, [14] the prime of Greek Epic precedes the attested phases of Greek Lyric by a considerable span of time and, what is more, features a highly complex meter of mysterious origins. The dactylic hexameter of those semi-prehistoric figures whom we cherish as Homer and Hesiod has no obvious formal antecedents. The Greek epic genre as such affords no clues, {5|6} since the only attested epic meter is the dactylic hexameter itself. In the end, Meillet was forced to leave out of his comparative study the oldest sample of Greek poetry (more than 30,000 Homeric and Hesiodic verses) on the grounds that the dactylic hexameter must have been borrowed—presumably from the civilization of the Mediterranean substratum. [15] This argument from silence has been tentatively challenged by Meillet’s follower Watkins, who points out the one obvious fact which suggests Indo-European provenience: the closing rhythm of the dactylic hexameter, – ⏑ ⏑ – ⏓, “is precisely that of the paroemiac line in Greek, as well as that of the South Slavic epic decasyllable.” [16] As both Jakobson [17] and Watkins [18] have argued convincingly, the closing paroemiac rhythm of certain Slavic and Greek meters is a basic feature of Indo-European verse. Thus at least the closing of the dactylic hexameter may indeed be Indo-European in origin. But the problem remains how to explain the overall shape of this verse in terms of the rhythm …– ⏑ ⏑ – ⏓.

If we could just find a way to support Meillet’s theory of Indo-European meter with the metrical evidence of Greek dactylic hexameter, many of the doubts harbored by Greek metricians would perhaps vanish. The task is arduous, because {6|7} we have to devise a plausible scheme for deriving the epic hexameter in terms of the Greek lyric meters which we already know, and all these meters are attested only in a period far later than that of the early Epic. Many Hellenists will find it difficult to think of the epic hexameter as a structure more recent than some lyric meters, simply because Greek Epic is known to be so much more archaic than Lyric in both diction and theme. Even a sceptic will admit, however, that an Aeolic meter like the expanded Pherecratic seems more primitive in terms of its internal structure than the hexameter: Besides the internal evidence, I will also adduce comparative evidence in the succeeding chapters to argue that the pher^3d is not just more primitive but more archaic as well. An experienced Greek metrician like Wilamowitz has intuited as much, and more: he suggests that a meter like the pher^3d had actually been the model of the hexameter. [20] True, he gives no specifics, nor does he define the pher^3d precisely, but the idea is there. Wilamowitz is by no means the {7|8} first, either, to consider the hexameter a highly evolved form which comes at the end of a long tradition. Theodor Bergk, the great editor of the Greek lyric corpus, was persuaded as early as 1854 [21] that the Iliad and Odyssey are not samples of metrical primitivism, but rather, the fullest realization of the potential that is inherent in Greek verse. [22] He makes a point of disagreeing with those who consider the dactylic hexameter the primordial type of Greek verse. [23]

My intent, to be sure, is not to argue that the epic hexameter was a new meter. I would say only that it was newer than some lyric meters such as the Aeolic. The hexameter is in fact archaic beyond our more casual imagination, as we may infer from the Parry-Lord theory of the formula. [24] Thanks to Parry’s brilliant discovery that the basis of the Greek epic language is a formulaic system, we now know that the Homeric (and Hesiodic) hexameter cannot be viewed simply {8|9} as an intricate rhythmical system (= meter) into which the poet fits his words. Rather, the epic hexameter consists of formulas which have fixed inner rhythms and which are joined together in such a way as to produce a fixed overall rhythm. If formula and meter are correlate phenomena, it is important that we be prepared to describe meter in terms of formula as well as vice versa. In this connection, we must notice the striking chronological variety that exists in the formulaic system of Greek Epic. For example, the formulas in one Homeric passage may contain two discrete points of information that can be dated as far apart as about half a millennium (from, say, the eighth to the thirteenth century B.C.). [25] Comparable chronological gaps exist between grammatical forms as well. [26] In short, the formulaic system of Homeric diction betrays a lengthy period of evolution spanning several centuries. If we allow for a parallel period of evolution in the meter which encases the formulas, it follows that the dactylic hexameter is an old meter. The argument remains, however, that lyric meters like the Aeolic are even older, from the comparative point of view.

In fact, I agree with the theory of Wilamowitz {9|10} that the meter of Greek Epic is derived from a meter of Lyric. Specifically, I will propose that the lyric meter in question is the same one which I have already described as the expanded Pherecratic (pher^3d), an Aeolic form which happens to display the closing rhythm … ⏑ ⏑ – ⏓ of the Indo-European ‘Paroemiac’. Besides my own arguments, which will be based on formulaic and metrical considerations, other arguments too are available in support of the general proposition that Greek Epic stemmed from Lyric. I cite in particular the ingenious observations of Pagliaro, whose reasoning is based more on content than on form. [27] Pagliaro uses as evidence those Homeric passages where the poet makes reference to poets and poetry as they exist within his narrative. [28] If indeed Homeric form and content span several centuries of evolution, it follows that references to poetry may betray different stages in the evolution of poetry itself. What Pagliaro finds is that the traditional terminology used in epic narrative for referring to the performance of Epic within the narrative reflects stages of evolution in genre which are far earlier than what we have as {10|11} the end result of such evolution, namely the Homeric corpus itself. And the more archaic the terminology, the more it seems to suit not only Epic but Lyric as well. Conversely, the more recent the terminology, the better it suits the Epic alone—specifically the Homeric end product. [29] If we adopt a teleological view of the Iliad and Odyssey as the culmination of a long tradition, then the intuitions of Giambattista Vico on Homer will prove to be more fruitful in this regard than the labors of early analysts like l’abbé d’Aubignac or even F. A. Wolf. [30] A unique epic form would have developed from multiple lyric forms within the social framework of performer-audience interaction. In the acme of this epic form, the reasoning goes, we should expect various stages of this evolution to be visible in the various references to the social framework. The point is, when Epic looks at Epic, it is forced to talk about itself in terms which are at times still suitable to Lyric as well.

Needless to say, the evolution of Greek Epic {11|12} is distinctive in content as well as form, and the medium itself underscores its awareness of its own distinction. For example, in the first book of the Odyssey, Telemachos refers to the bard Phemios’s treatment of the theme of The Return of the Achaeans (ʼΑχαιῶν νόcτον: α 326) as a ‘song’ which is the ‘newest’ and therefore popular with audiences (α 351f):

Inside the art of narrative, of course, ‘newest’ implies that the ‘song’ treats the recent events after Troy was sacked. Outside the art of narrative, however, ‘newest’ may refer to the genre which treats the Return of Odysseus, namely, a genre within the Odyssey itself. It is described as popular by virtue of being a recent song (α 351f). I believe that the poet of the Odyssey is here making a self-conscious reference to his own genre, or even to his composition, by means of an adjective which means one thing in the narrative of his composition but another thing in the reality of his performance of the composition. [31] {12|13}

Even if Greek Epic is a newer form than Lyric, the attested archaic meters of Lyric fail to be associated with a formulaic system of the sort that characterizes Epic. Since the evidence of Homeric Epic conditions us to think of formula as a correlate of meter, this apparent lack of a formulaic system in Lyric causes some worry. Are we to think of the surviving lyric meters as archaic rhythms without archaic diction? Are they empty shells from the standpoint of traditional phraseology? The extenuating factor here is attestation. By the time Greek Lyric is actually attested, even the dynamic art of Epic—let us call it oral composition—is already moribund if not dead. [32] Once the formulaic system breaks down, it is no longer possible to compose while you recite and recite while you compose. [33] Yet the epic genre by no means loses its traditional phraseology overnight, as it {13|14} were. The diction of a distinctly non-oral production like the mock-epic Battle of Frogs and Mice, [34] which may well be roughly contemporaneous with the lyrics of Sappho, is essentially parallel with Homeric diction. [35] Thus I would argue that even if the lyric diction of someone like Sappho may not be strictly formulaic, it could still depend heavily on earlier formulaic systems that had been appropriate to her archaic meters. In her composition known as The Wedding of Hektor and Andromache, which is the largest piece of Sapphic narrative so far attested, there are numerous traces of traditional diction which is not only appropriate to her traditional lyric meter but also independent of epic hexameter. [36] Among traditional expressions in this poem, we find κλέοc ἄφθιτον.

Let us return to the one Greek meter where we can still see formulas at work, the epic hexameter. In particular, let me mention an interesting side-discovery made by Parry, who noticed that the traditional caesuras and diaereses of Homeric hexameter mark the places where formulas start or stop. [37] When I tried the {14|15} approach of isolating the shapes of the epic formulas (as they are marked off by the ends of the verse and by the main word-breaks), I discovered that they matched the shapes of certain basic meters of Greek Lyric. I reserve the details for Part I, but I must stress even now the significance of such a formal convergence. If Meillet had looked at the shapes of the Homeric formulas rather than the shape of the overall hexameter, he would have seen a direct formal link with the oldest Greek lyric meters. What had been part of the problem for Meillet may yet be turned into part of the solution. The epic hexameter provides valuable comparative evidence not only from the shape of its formulas but also from their function. In this regard, there is one Homeric formula which will be of particular interest for our study, that is, κλέοc ἄφθιτον (ἔcται) in Iliad I 413.

Once we are in the position to define the metrical context of κλέοc ἄφθιτον in terms of Greek Lyric and Epic combined, we will have to compare it with the metrical context of śráva(s) ákṣitam in the Rig-Veda. Here the strictly formulaic considerations found in the Greek portion of our study cannot be applied directly. In fact, I will refrain from ever using the word ‘formula’ in referring directly to Rig-Vedic phraseology, despite its impressive archaism and internal consistency. The edifice of the Rig-Vedic hymns is built on a formidable set of mnemonic devices designed for the transmission {15|16} of old fixed texts over immense stretches of time. I have little doubt that some layers of the primordial texts were once oral compositions, but we have to reckon with the factor of transmission. If we use Lord’s sensible criteria, [38] a truly ‘oral’ tradition is one in which every performance generates a new composition. The factor which remains a constant is the singer’s repertory of formulas. The Vedic genre operates on a different and far stricter principle. The Vedas are a vast collection of hymns devolved from prayers, and prayers are a genre which encourages fixed performances tailored for each sacral occasion or need. In short, we are dealing with an art-form grounded in religion. [39] The {16|17} ostensible audience is divine, not human, so that even human comprehension is not a prime consideration. Words in a prayer might no longer be understood by us mortals, but the prayer remains efficacious and so the words must be right. [40] As we compose new prayers, we will re-use incomprehensible elements even at the risk of distorting their original meaning or grammar. Granted, secular poets also use incomprehensible traditional elements, but they cannot keep their audience if they go beyond mere flavoring of their diction with arcane language. There is no such compunction in Vedic ritual, where arcane language is a precious heirloom which keeps on inspiring language that is even more arcane. [41] My point remains that the Rig-Vedic hymns are a chronologically varied set of fixed texts inspired by other fixed texts, all of them transmitted orally over many centuries. [42]

Since the very format of the Rig-Veda nurtures {17|18} archaism for the sake of archaism, the concentration of its archaic phraseology and the regulation in the patterns of this phraseology come as no surprise. This extreme strictness should be evident to anyone who takes the trouble of reading through Chapter 9, which maps out the distribution of the word śrávas in terms of the Rig-Vedic verse. The licenses in the rhythmical patterns of the verse belie the strictness in the distribution of the phraseology. If I were allowed to ignore time and space in my comparisons, I would certainly cite Gerard Manley Hopkins: “So I may say my apparent licenses are counterbalanced, and more, by my strictness.” [43] As we examine the relationship of Rig-Vedic phraseology with Rig-Vedic meter, we find that the traditional phrase śráva(s) ákṣitam occurs in a formal context which matches that of κλέοc ἄφθιτον in Greek Lyric and (indirectly) in Epic. To my mind, such a metrical and phraseological convergence is extremely important because it suggests that we have here the remnant of an Indo-European poetic phrase. With the comparative evidence of this Indic phrase śráva(s) ákṣitam and the corresponding Greek phrase κλέοc ἄφθιτον, I will attempt to draw up an overall theory of formula and meter in Greek versification (Chapter 6).

There might be some who worry about the possibility of coincidence. Could it be that the {18|19} adjectives ἄφθιτον and ákṣitam just happen to be chosen for the nouns κλέοc and śrávas, so that all we have is an accidental convergence of cognates? In the Epilogue, I have devised a counter-argument inspired by Parry’s discoveries about the function of fixed epithets. To put it simply, if we survey all the contexts of ἄφθιτο- and ákṣita-, we see an overarching function in this epithet which is singularly appropriate to the noun which it describes. We will see that the epithet + noun combination meant something special and that this special meaning is essentially the same in both Greek and Indic. Form and content match in my reconstruction of *klewos n̥dhg^whitom.

Before I plunge into the specifics, I should outline those basic works and ideas which I have found most helpful in my approach to Greek and Indic meter. First of all, I rely heavily on the Parry-Lord theory of the formula and on the techniques of formulaic analysis, [44] including those of Hainsworth, Hoekstra, and Edwards. [45] Secondly, I strive to follow Meillet’s methods of comparative reconstruction, with which he has been so successful in metrics as well as linguistics. [46] I have already mentioned two other important comparative approaches to meter, those {19|20} of Jakobson and Watkins. [47] As for internal analysis and reconstruction, I should single out Wilamowitz, Snell, Maas, Allen (Greek metrics), Oldenberg, Arnold (Indic metrics). [48] For general metrics and prosody, I have often consulted the useful work of La Drière. [49] As for the technical but essential terms related to meter, I have tried to abide by the norms of Webster’s Third International Dictionary. I hope to specify as I go along wherever I deviate from the standard nomenclature.

Already at this point, I should note that I will use two idiosyncratic terms which call for particular attention because of their importance for certain stages of my approach. I owe to Saussure the concept of distinguishing the synchronic from the diachronic. [50] From the synchronic point of view, we see a structure as it exists in a concrete time and space; from the diachronic point of view, we see it evolving in the abstract. For my very first application of these terms, I take this opportunity to cite a general definition of meter which to my mind merits special consideration: "The metrical pattern is a copy, a mimicry, a counterfeit without intention to deceive, of {20|21} the basic elements of our language and of their order. When this metrical pattern and a set of words or phrases are placed in conjunction, a tension exists between them, a strained state of mutual relations; and the language, when we read it as part of a poem, has been strained into something different, into a resemblance to an imitation. It is thus partly imitation itself, and it is this which makes it an art." [51] My own work on meter seeks to add another dimension to this definition by the application of a synchronic/diachronic perspective. I suggest that meter and its poetic language are indeed an imitation of natural language on the synchronic level, but that they can be an actual outgrowth of natural language on the diachronic level. Throughout my monograph, I will seek opportunities to support this claim with evidence from early (even preliterate) phases of Greek and Indic versification.

Mention of claims and evidence brings me to the final point of my introduction. I wish to be the first to admit that I need and seek more evidence for the theories which I offer. And yet, obviously, the immediate availability of evidence is a factor to consider. On one extreme, I have been able to adduce a multitude of data, {21|22} not just the example of κλέοc ἄφθιτον (ἔcται), for my theories on formula vs. meter and on the origins of Greek hexameter. On the other extreme, the theory that cognate poetic phraseology matches cognate meters in Indo-European languages is here supported mainly by one set of comparanda, Greek κλέοc ἄφθιτον and Indic śráva(s) ákṣitam. This particular restriction to one set, however, is not so much for want of other possible examples as for reasons of scope. At the present stage of research, I submit, monographlength inquiries are needed to explore adequately the cognate metrical features that may be latent in each set of cognate poetic phraseology. [52] With the help of more evidence, to be sure, I can expect to revise many of my hypotheses and perhaps also renounce some. For the time being, however, I feel eager to share my own findings up to now, such as they are. Perhaps it is at this point that I should also make a comment on my mode of presentation. In instances where I worked especially hard on the wording of a given hypothesis, I recorded my formulation in italics. {22|23} I did so not to express any confidence about solutions, but rather, to suggest the importance of the problem under consideration. I have similar feelings about the overall intent of my work, and I can only hope that the reader will be indulgent and interested.