One
Traditional Oral Poetics
This chapter, like the volume in general, explores some fundamental issues in oral literature research from a comparative point of view. It consists of three parts: (1) a brief examination of certain of the most significant contemporary problems in oral traditional poetics, (2) a "program for reading" traditional texts (as opposed to a poetics intended for literary works), and (3) a sample application of the program to five texts from the Serbo-Croatian, ancient Greek, and Old English poetic canons. While I cannot claim to confront all of the relevant issues in this single chapter, I hope that these initial remarks have meaning for those who aim at an understanding of oral and traditional texts on their own terms. For that is precisely the legacy and challenge of oral literature research to date: namely, to attempt the untried trajectory from structural explanation to a unified poetics for these texts. We cannot, nor in fact do we need to, settle any longer for the comfortable sinecure of a strictly literary criticism; we need to strike out, to make bold new hypotheses based on the solid data of recent discoveries about oral and traditional poetry, for these hypotheses are bound to be closer approximations than those based on strictly post-oral, non-traditional verse. In a preliminary way, then, this opening chapter tries to respond to this need for a new poetics.[1]
Contemporary Issues
The first and foremost area of uncertainty in oral literature research has to do directly with two key words—oral and traditional —as they were first used
[1] I shall devote a future study, tentatively titled Immanent Art , exclusively to the directions mapped out in section 2 of this first chapter, "A Program for Reading Traditional Texts." Since, however, the implementation of the reading program can proceed only after a firm foundation in the comparative philology of oral epic poetry exists, the present volume will concentrate on the various levels of structure in comparative oral epic.
by the classicist Milman Parry in referring to Homer's Iliad and Odyssey .[2] To take the latter first, note what Parry wrote in 1928 about the nature of Homeric language:
To establish in the Iliad and the Odyssey the existence of an artificial language is to prove that Homeric style, in so far as it makes use of that language, is traditional. For the character of that language reveals that it is a work beyond the powers of a single man, or even of a single generation of poets; consequently we know that we are in the presence of a stylistic element which is the product of a tradition and which every bard of Homer's time must have used. (1928a, in 1971, 6)
Parry is describing a formulaic language,[3] a diction which from our text-centered point of view is repetitive or recurrent, and he is deriving from the existence of that diction a poetic tradition . He does not mean by his use of that word what T. S. Eliot meant in his famous essay on what is finally an unrelated topic, nor does he mean (as he has often been accused of meaning) that Homer is trapped or conditioned by his language. The poetic tradition properly understood is not at all a limiting but rather a connotatively explosive medium, a touchstone or nexus of indication and reference wholly different from the medium at the disposal of the "non-traditional" artist, for such a diction and narrative structure have obvious and necessary reference not only to the present poem, poet, and time but also to an enormous number of other poets, poems, and eras. Written diction and tradition, no matter how dense with allusions and inherited figures, can never command this open connotative field of reference.[4]
Scholars reacting to Parry's original writings and to Albert Lord's later and equally masterful scholarship have seldom understood this special sense of traditional . They have too often perceived the language and narrative weave of Homeric epic and other such works as closed systems, that is, as fully formed, mechanical complexes which the poet need only activate in order to re-create the poem. Analyses of traditional patterning have, in the great majority of instances, implicitly assumed a synchronic text, one which need merely be formulaically, narratively, or syntactically parsed in order to be explained and faithfully interpreted. What scholars have rarely recognized is
[2] In this volume there will be no attempt to refer to a wide or even representative selection of general works on oral literature research. For a historical introduction to the field, see my Theory of Oral Composition: History and Methodology (1988); for bibliography through 1982, see Foley 1985, updated annually in Oral Tradition .
[3] Parry's original definition of the formula (1930, in 1971, 272) is "a group of words which is regularly employed under the same metrical conditions to express a given essential idea." Lord's definition of themes (1960, 68) is "the groups of ideas regularly used in telling a tale in the formulaic style of traditional song." For references to further work in more than ninety separate language traditions, see Foley 1985 and note 2 above.
[4] See further, e.g., Renoir 1981b and Foley 1986a,b.
that a traditional text is not simply a synchronic latticework, but also a diachronic document of great age and depth. For tradition is nothing if not diachronic: it has roots which reach back into its pre-textual history and which inform the present avatar of its identity. Just as, given the insights of recent literary criticism, we cannot any longer settle for one-dimensional models of Jane Eyre or the works of Proust, so we cannot afford to countenance the reductive distortion of a poetic tradition to a closed synchronic system. We must actively affirm the historical and evolutionary nature of oral tradition, for this is a crucial aspect of its context.[5]
The variously construed term oral is nearly as important as traditional to our grasp of poetics. Should we agree with Parry and Lord in carefully limiting the textual sample to oral epic (usually from the Moslem rather than the Christian tradition) composed by entirely unlettered singers, or should we broaden our perspective—and if so, how much should it be broadened? The spirit of Ruth Finnegan's (1977) extension of the corpus of oral poetry beyond the epics of Yugoslavia, ancient Greece, and elsewhere to the myriad other genres composed and performed orally around the world can only mark a positive—even necessary—development, since we need to understand the larger context of oral poetry as well as the material on which we choose to focus. As I have tried to illustrate elsewhere (Foley 1980a), cross-cultural comparisons can quite usefully be drawn between members of genres other than epic; the Old English and Serbo-Croatian magical charms provide one opportunity for such an examination. What must be observed rigorously in every comparative undertaking, however, is the integrity of genre. One simply cannot expect a cogent analysis to come out of a comparison of, for example, riddles and epics; the generic assumptions implicit in the forms must be at variance, and this variance seriously reduces, if not actually invalidates, the legitimacy of the proposed comparison. If we wish to include a number of genres in our survey of an oral culture, we need to proceed in our analyses along methodological lines which respect the principle of genre-dependence .
The term oral also begs the question of what is to be included under its heading. Not only various genres but also various kinds of texts present themselves. Some of these, like the Serbo-Croatian epske pjesme recorded by Parry, and Lord,[6] arc known to be oral in origin and in execution; they contain no ambiguous elements possibly traceable to written composition and can confidently be classed as oral traditional texts in the truest sense of that rubric. Another kind of text, much more common and much more commented upon, is that which has come down to us in manuscript only. We may have a few hints about its recording, but these clues are often undependable or mutually contradictory. Often this second kind of text suffers from further complications,
[5] See esp. G. Nagy 1974, 1979; Peabody 1975; Janko 1982.
[6] For a description of the collecting, see SCHS 1:3-20; Lord 1951b.
such as the idiosyncrasy of its recording or editing (one thinks of the stitched-together Kalevala of Finland) or the fragmentary or multilingual nature of its remains (the Gilgamesh ). In light of the controversy over the formulaic test for orality and its frankly uneven application and results,[7] which of these manuscript texts do we admit to fellowship with known oral material, and which do we exclude?
Again I would recommend making some finer distinctions than are customarily pursued by those on either side of the controversy. The formulaic test as it has generally been carried out cannot prove oral provenance,[8] for as long as scholars commit the egregious philological sin of importing models and definitions directly from ancient Greek to other poetries without taking account of necessary differences in prosody and versification, nothing can be proved. By counterposing Homeric phraseology to the diction of Old English, Old French, or whatever other poetry one chooses as comparand, without making adjustment for the individual characteristics of each poetry, one simply calculates the extent to which the compared work is composed of Homeric Greek formulas, obviously a useless index. This simple principle of tradition-dependence , that is, of respect for a given literature's linguistic and prosodic integrity, has consistently been ignored by comparatists eager to achieve what may seem like ground-breaking results but which are actually based on flawed assumptions from the start. Along with genre-dependence, this attention to language-dependent features would clear away much of the confusion which presently impedes progress in oral literature research.
If the formulaic test for orality must be abandoned (at least in its present form),[9] then what can we do about categorizing our manuscript texts of uncertain provenance? Again I find the question as contemporarily phrased too exclusive and heavy-handed, too much like the Higher Critical battles of the nineteenth and early twentieth centuries over the riddle of the Homeric Question. If we cannot in the present state of knowledge confidently pronounce a problematic text oral, then why not admit that inability and proceed from there? Scholars have shown us that the Chanson de Roland and Beowulf , for example, have oral traditional characteristics : both texts demonstrate a formulaic phraseology, an inventory of typical scenes, and so on.[10] If that information is not enough to prove beyond doubt their orality in the specific Parry-Lord sense of a guslar singing in a kafana , then so be it; certainly the demonstration of oral traditional characteristics is not entirely in vain.
[7] As noted in Bynum 1978, 10-11. For the complex history of formulaic studies in Old English, see Foley 1981b, 27-122; Olsen 1986, 1988.
[8] See now Lord's similar view (1986c).
[9] See chapter 3 for a detailed exposition of how deep the prosodic differences are among the ancient Greek hexameter, the Serbo-Croatian deseterac , and the Old English alliterative line.
[10] See J. Duggan 1973; and, in general, Foley 1985. On the controversy over what such demerits mean to the provenance of a text, see, e.g., Duggan 1981a,b versus Calin 1981a,b.
But it is precisely this faulty conclusion—that is, that because a specific formula count or a certain kind of narrative recurrency is lacking we can cast a text back into the comfortable confines of a written, non-traditional literature—that constitutes a distortion fully as culpable as the overly ambitious assertion of orality that it seeks to correct. For it does not follow that tradition, even oral tradition, ends with the poet's or culture's first draught of literacy. What does end, unambiguously, is the oral tradition in its Ur-form, together with the possibility of recording the oral text in its Ur-form. What continues, just as unambiguously, even in the hands of a skilled writer of verse drawing on an earlier mode, is some vestige of orality and some vestige of tradition.[11] From this perspective, texts which exhibit undeniably oral traditional features, no matter how uncertain a provenance a fair examination of their known history may produce, cannot be treated as or classed with literary works of a much later time.[12] If we must then leave some of our manuscripts in an intermediate category, let us make sure that we opt for that more accurate (if finally indeterminate) characterization rather than settling for the apparently simple but imprecise model of "oral versus written" texts.
A Program for Reading Traditional Texts
In this section I shall gather up some of the points made above and attempt a brief codification. In the process it should be possible to posit a short program for confronting both unambiguously oral texts , such as the songs collected by Parry and Lord, and what I call oral-derived texts , that is, the manuscript or tablet works of finally uncertain provenance that nonetheless show oral traditional characteristics. This methodology must of course be preliminary and suggestive, for at this relatively early stage in the development of oral literature research no program can claim to offer more than a starting point for the poetics of oral and traditional works. But it can from a general perspective draw a few fundamental distinctions about the ontology of different kinds of texts, distinctions which have not been made and very much need to be made. What is more, even these first few steps can help to point the way toward the controlled, exacting comparison to be attempted in the succeeding chapters of these studies.
First is the matter of the term text , a concern purposely left undiscussed in my opening remarks. In using this term I mean to indicate a real, objective, and tangible score, an entity that exists both as a thing in itself and as a directive for its perceivers. In this second sense I take advantage of current
[11] See esp. Ong 1967, 1977, also 1982. Recently Lord has presented new findings on the influence of fixed texts and transitional material (1986a,c). Also worthy of note in this regard are Opland 1975, 1983; Stock 1983.
[12] Compare the remarks of Renoir 1986 on Sir Gawain and the Green Knight , Cynewulf, and the Hildebrandslied .
critical notions about the "activity" of a text, the dynamics or chemistry of its parts when brought together and to life by the reader.[13] Both senses of the word are important, as we shall recognize explicitly later on: the text serves as an object and as a libretto for the reader's or listener's personal "performance" of the work.
In addition to the dual nature of the text as objective and subjective, I would also stress its special identity in relation to the oral tradition from which it derives, whatever the exact tenor of that relationship might be. For example, a "pure oral text," as described above, represents only one performance of a given song and song-type by a given singer, so that the generic levels of the performance multiply rapidly as the context is determined (see Lord 1960, chap. 5). These levels of generic form are the lifeblood of the tradition, for they insure its recurrency over time and over place. Using the word and concept text allows us to designate the particular object we desire to confront or analyze within this reduplicative series. Likewise, in the case of an oral-derived work, we can again specify the one version of a multiform to be confronted, or, if the manuscript poem in question is too far removed from the tradition to be simply one instance of a multiform, the term text can still designate the unitary document which in this case will be comprised of quite different elements. As we shall see, this final instance proves to be a significant one, for when a poem is informed by both a traditional and a non-traditional, learned aesthetic, there is very often a tendency on the part of critics to eliminate either the one or the other aspect.[14] In an effort to make its poetics more straightforward, these critics would dismiss the vigor of such a hybrid text by miscasting its real nature or by characterizing it pejoratively as a "mixed" work. In this case the label text helps to maintain the centrifugal parts of the piece for resolution by a centripetal poetics.
A second step in our reading program would be the determination of whether a given text is actually oral or is in fact oral-derived. In either case, of course, we assume the work in question to be at least in part traditional because it exhibits unambiguously oral traditional characteristics. Further, and this point should be stressed, the minimal assumption of an oral-derived text does not automatically preclude true orality; it merely admits a degree of uncertainty about textual history and adopts a demonstrably valid position in favor of a likely conjecture.
For example, the Milman Parry Collection houses thousands of known oral epics, some of them recorded from guslari acoustically while sung, others recorded while spoken without the accompaniment of the single-stringed gusle , and still others taken from dictation at a much slower pace. As Lord (1953) has demonstrated, these kinds of texts are somewhat different in extent and
[13] See, e.g., Foley 1984b and 1986b.
[14] Some particularly well conceived exceptions to this rule are Renoir 1986, 1988; Olsen 1984.
detail, though usually not significantly different in general structure. In the nineteenth century, however, before modern acoustical recording apparatus became available, Serbo-Croatian epics were committed to a fossilized textual form exclusively through the agency of a scribe or scribe-editor.[15] In such a case we would want to know as far as possible the scribe's control over the performance and his subsequent or concurrent editing of the manuscript. In short, whatever intervenes between the spoken word and its apotheosis as a written record, whether during the performance or at some later time, contributes to the history of that oral text as part of its context. To settle for the most convenient (because most visible) stage of that process—the printed edition alone—is to misread the oral text.
Should we on the other hand be confronting a manuscript work whose certain orality we cannot prove, many of the same criteria for judgment and meaningful reading still apply. If we know only a few details about the recording or (let us say, in order to avoid ambiguity) the creation of a given text, we must nonetheless take them into account; while these few details may fall short of a complete description, they do contribute to context. In the case of an oral-derived text, many of these features will be paleographical in nature. For example, do the graphemics of the manuscript indicate anything about its character? Or does the codex which contains it offer a reading context of any sort? Perhaps we may be able to unearth some information about the manuscript history of such a text—for instance, about the vexed question of the Alexandrians' treatment of Homer. Again, as in the case of the certainly oral work, the oral-derived text demands as its context all that we can discover about what intervened between its formation as a text and the copy in our hands.
The judgment of either oral or oral-derived tells us in general terms what is to be expected in a given work. In the former case we can expect, with certain qualifications resulting from the circumstances of creation and transmission discussed above, a text whose poetics follow characteristic outlines best determined by an intimate acquaintance with the tradition from which the text takes its meaning. Reading as many works as possible within the tradition—that is, developing what synchronic linguists call a "competence" or ability to understand the "traditional grammar"—provides the most thorough preparation. With a known oral poem and a well-collected oral tradition, one can approach this ideal situation and develop an awareness of a work's traditional context.[16] This experience will go a long way toward the formulation of a poetics for that text and tradition, and it will also provide the best possible
[15] For a general survey of this process and its fruits, see Coote 1978. Cf. the insights of the recent study of performance texts by Fine (1985).
[16] On using a computerized text-processor to "read" numerous versions of a song concurrently and thereby to gain a sense of the gestalt that is their tradition, see Foley 1981a, 1984a.
preparation for comparative work, especially in that large number of cases in which the literature to be compared is not certainly oral or well collected.
And, of course, not only are many known oral traditions not well collected, but almost all manuscript poetries with oral traditional features are very sparsely recorded.[17] In such cases, in which the most immediate context is lacking, one would do well to supplement the investigation with a comparative context, seeking to elucidate one literature through an analog.[18] This was in the early going Parry's and Lord's basic methodology, turning the observable realities of Serbo-Croatian oral epic back on Homer to illuminate the ancient Greek texts by comparison. In the instance of at least some of the Old English and Old French poems, however, such a comparison would obscure the differences between oral and oral-derived. As a corollary to the Parry-Lord procedure, then, I would suggest that scholars in search of analogs for oral or oral-derived texts look first to possibilities in the respective categories. It may turn out, as in these studies, that the criteria of tradition- and genre-depen-dence make comparison between and among texts from different categories especially attractive and promising, and that these primary considerations must override the secondary matter of oral versus oral-derived. But no matter what the particular situation, the object remains the most exacting, controlled, and productive comparison possible.
A third procedure in our program of poetics is the observation of the principle earlier termed genre-dependence . Many potentially fruitful comparisons have been to some extent qualified by drawing analogies between very different genres, such as between lyric (and non-narrative) panegyric and narrative epic. As a first approximation, genres in two or more literatures should be either narrative or non-narrative; that is, they should be either forms that involve the progress of a tale, such as is found in epic, or the kinds of momentary, brief song-types such as lyric which have no narrative dimension to speak of. More specifically, the works to be compared should be, as far as is feasible in separate poetic traditions, precisely the same genre. Of course, this sort of exact congruency is not always practically achievable, and even when it may seem from formal criteria that we have a close fit, often enough that is not really the case;[19] nevertheless, by aspiring to fulfillment of the criterion of genre-dependence, we engage a given poem in as appropriate a cross-
[17] Perhaps the sole Western European exception to this rule of thumb is Old French epic poetry, the chanson de geste tradition, on which see esp. J. Duggan 1973.
[18] As Renoir (1981b, 439-40) envisions it, such comparative criticism is advantageous in two respects: "The first is that it provides a flexible but specific context whose integrity is readily ascertainable and against which we may in turn measure the integrity of a given text, about which we are thus enabled to make reasonably objective statements. The second is that it provides us with the means for understanding certain aspects of ancient texts through legitimate comparison with more modern texts."
[19] On this point, see Ben-Amos 1969; Dundes 1962.
traditional comparison as possible. In turn, the appropriateness of the comparison, and therefore of the object poem's comparative context, will help us to formulate its poetics most sensitively and to improve our competence as readers of the poem.
A fourth programmatic principle of approaching an oral or oral-derived text, that of tradition-dependence , itself consists of three smaller parts. In general we can say, as was remarked above, that literatures differ significantly from one another, most obviously in their linguistic properties but also, because they are dependent on linguistic realities, in their prosodic properties. A viable poetics must take account of these idiosyncrasies in assembling its textual profile. As principle 1, then, I would set as the first goal of tradition-dependence a close philological mastery of the tradition within which the given text emerges.[20] The rigorous understanding of the text in its own language and its most immediate context would then make principle 2, a meticulous survey of differences as well as similarities between or among a text and its analog(s), a credible endeavor. Without both principles 1 and 2 we fall victim to the superficial gesture of the dilettante, uncritically remarking similarities at the expense of a faithful overall perspective. While this kind of rigor is certainly necessary for a successful reading of a known oral text (otherwise a false context is established), it is even more crucial to understanding the poetics of an oral-derived text, where the traditional features to be understood are not as unambiguous, and perhaps not as clear or as frequent, as those in the known oral text.
Principle 3 pertains mostly to the certainly oral work, but may as an abstraction affect the manuscript poem as well. It concerns the place of a given text within what may be called the "local tradition" (as opposed to the national tradition as a whole) and within the repertoire of the given singer. Again the Parry texts provide an illustration. As one becomes more and more acquainted with the Serbo-Croatian tradition, he senses increasingly the truth of Lord's caution that one must read the songs of a given singer and of his fellow performers in order to understand the local tradition and the individual guslar's role in it.[21] Even though the poetic language of any region is multidialectal and "polychronic" (cf. Parry 1932 on Homeric Greek), containing archaisms alongside contemporary diction, it will nonetheless bear the characteristic imprints of its locale and also of its user's idiolect. The epic language at large is like a grand inventory of smaller, specialized sets of local diction, and these
[20] Although this might seem the most obvious prerequisite for any sort of study of literature, in fact comparative analysis of oral traditions has suffered from lack of its fulfillment. Those who summon the Serbo-Croatian tradition as an analog without considering anything more than translations are among the most common, and most egregious, offenders.
[21] For illustration of the three levels of tradition (the individual or idiolectal, the local or dialectal, and the national), see chapters 5 and 8 on, respectively, formulaic and thematic structure in the Return Song from the Stolac region.
sets are in turn divided into yet more specialized individual sets whose character depends on geography, religious affiliation, repertoire, and the influence of other singers. All of these factors contextualize the text by being aspects of its tradition-dependence.
The fifth and last theorem in our brief program is the question of the confluence or co-existence of synchronic and diachronic forms in the single traditional text. We may speak generally of textual activity on the synchronic level by describing a given text's digest of formulas, themes, sound-patterns, story-patterns, and the like. For the known oral text these elements have their reality in the act of text-making as it is carried on by the poet, the nominal author of the work and the particular author of the text at hand. Further study of his repertoire and of the community of singers in which he functions will create a synchronic context, a profile for the text. But it would be shortsighted to judge the process of contextualization complete with this profile. As indicated in the opening section, a text has also a diachronic context, a historical or evolutionary aspect which extends far back into the pre-textual history of a given work.
We know that tradition preserves what is of value to it from the past, and we also know that that preservation is not a consciously designed undertaking but rather a reflex of the tradition itself.[22] For example, archaeologists have pointed out the chronological strata in Homer (e.g., Shodgrass 1971); however, the poem apparently senses no flaw in what we would see as an anachronistic juxtaposition of elements or beliefs. From another point of view, Gregory Nagy (1974) has shown that the roots of the Homeric formula kleos aphthiton are, along with its Sanskrit cognate sráva(s) áksitam , to be found in Indo-European mythology; again it is diachrony, in this case of great depth, that fills out an important part of the poem's context.[23] In fact, of course, these two examples have reference to materials and ideas that are anachronistic only from our decidedly synchronic perspective of post-oral and post-traditional thinking. Tradition simply finds no discrepancy here; elements from different eras, even vastly different eras, are held together in a diachronic suspension unique to traditional texts. We need to take more careful account of this historical or evolutionary perspective, even when its traces are difficult to interpret either because they exist only as fragments or suggestions or because we must ferret out new information; and we need to look for its contextualizing influence in both oral and oral-derived texts.
Application of The Program: Approaches to Reading Five Traditional Texts
It is time now to put the reading program into action, to test its power to elucidate some actual texts. In doing so I am aware of the objection that a
[22] On the "homeostatic" nature of oral tradition, see, variously, Peabody 1975; Turner 1986.
[23] For other examples of diachronic approaches, see Hoekstra, e.g. 1964, 1969; Janko 1982.
poetics is not possible until the structure and dynamics of the texts have been established. What I hope to accomplish here is merely a prolegomenon to a subsequent work that will be based on the structural principles with which this volume is concerned. Nevertheless, I do feel that it is important to know where the entire investigation is headed from the start; the applications made below are thus intended simply to indicate some profitable directions for future criticism, and certainly not to exhaust their possibilities.
What I propose to do, then, before turning to the oral and oral-derived epics that are the real subject of these studies, is to try out the reading program on five traditional texts or groups of texts: (1) a Serbo-Croatian epic; (2) the Homeric Odyssey ; (3) Serbo-Croatian and Old English magical charms; (4) the Old English Beowulf ; and (5) the shorter Old English poem The Seafarer . Our steps may be outlined as follows:
A. Question of "text"
B. Oral or oral-derived
C. Genre-dependence
D. Tradition-dependence
1. Original language and philology
2. Comparison/contrast with other traditions
3. National, local, idiolectal levels
E. Synchronic and diachronic contexts
In this way I hope to present by example the outlines of a traditional poetics in an applied form, a poetics flexible enough both to accommodate the various kinds of texts one can and does encounter and at the same time to allow the necessary focus and refinement that each individual tradition, singer, and text requires and deserves.
1. As my first example, a Serbo-Croatian oral epic, I choose an easily accessible published text, The Captivity of Djulic[*] Ibrahim (Ropstvo Djulic[*] Ibrahima ), Parry no. 674 sung by Salih Ugljanin on November 24, 1934, in Novi Pazar and published as no. 4 of the first two volumes of SCHS (original-language text and translation, respectively).[24] Because the primary form of the song is that of an acoustic recording, a performed text whose sounds are recoverable, the objective and subjective aspects of the work come as dose as possible to superimposition. While we cannot re-create the original setting and audience, we can re-create much of the rest of the performance and, to the extent that we are ourselves acquainted with the tradition, become an audience in an approximate way.[25] So the question of text proves in this way to be an easy
[24] I use this particular version of a Return Song because of its ready access in both the original and translation; as documented below, virtually all of the material analyzed in these studies also belongs to the epic subgenre of Return. See esp. chapters 5, 8, and 10. On the standard Parry Collection notation, see n. 51 to chapter 2.
[25] Compare the observations on text and audience that are interwoven throughout the essays in Tedlock 1983.
one, as long as we do not confuse the published edition with the performance as preserved on aluminum records; by recognizing the real text as a sung and heard reality informing the edition we read, we can proceed to the next and subsequent steps prepared for comparison. We shall find later that the present working designation of Ugljanin's text—as Ropstvo , version no. 4—will bear emendation to identify its context, but at this point the indication of a unitary document is entirely appropriate.
Nor do we need to pause long over the question of oral versus oral-derived: the Ropstvo text is a known oral epic and is therefore best contextualized by a thorough knowledge of Serbo-Croatian oral epic tradition. In this particular case such an acquaintance can be readily gained by consulting the resources of the Milman Parry Collection and its many hundreds of epic texts, and secondarily by reference to other collections of Serbo-Croatian oral narrative (e.g., Karadzic[*] , Marjanovic[*] , and Hörmann). Most immediately, this text was sung to the accompaniment of the gusle (rather than spoken for the records or taken from dictation); it will thus relate in certain formal aspects to other sung texts.[26] Furthermore, it was composed by Ugljanin, a considerable part of whose repertoire has been recorded and published in the Serbo-Croatian Heroic Songs series. This material forms another sort of context, which may be measured comparatively for various stylistic properties—such as formula, theme, frequency and kind of enjambement, and so forth—and so used to deepen our understanding of Ropstvo no. 4. The most obvious analogs in this regard are nos. 5 and 6—recited and dictated versions, respectively, of the Ropstvo .[27] Widening the textual circle a bit more, we can then proceed from compositional mode and repertoire to an analysis of the local tradition. Here the comparative method extends the field of inquiry to the songs and repertoires of other singers, with comparisons and contrasts including all characteristics of composition and other performance data through the shape and function of large narrative patterns. The known oral text may thus command a host of contexts, all of which are important to reading the text and thus constitute part of its poetics.
After these determinations we shall want to establish the generic context of Ropstvo no. 4, so that worthwhile comments and comparisons can be promoted and unproductive ones avoided. The Ropstvo is an example of a relatively well known subspecies of South Slavic and Balkan epic, what Lord has called the "Return Song."[28] To proceed from general to specific, then, the text in question can be classed as "epic," with all the generic connotations that accompany the term in Serbo-Croatian poetic tradition. We may summon for comparison
[26] See Lord 1953, and chapters 5 and 8 below.
[27] See Lord's comparisons in SCHS 1:339-58.
[28] On the pattern of Return, see esp. Lord 1969; Foley 1978c, 1986b. For a full list of relevant scholarship, see chapter 10 below.
all types of oral epics within the tradition and perhaps also from related and unrelated oral traditions, being careful to weigh the worth of each proposed comparand according to the requirements of the reading program, and thus establish for the object text a wide but finely differentiated context. After that initial step, the specific nature of the Return Song and its various avatars in the Yugoslav and other traditions should be addressed. This process will include relating the Ropstvo to the story-pattern of Return, abstracted by Lord (1969) in the form A—D—R—Rt—W, where A = Absence, D = Devastation, R = Return, Rt = Retribution, and W = Wedding. The characteristic form of this kind of epic entails a hero long absent from his homeland (A); while he suffers in captivity, suitors consume his possessions and attempt to marry his fiancée or wife (D); he then returns (R) and takes his revenge (Rt) on all those who have plotted against him; and finally, if he finds her faithful, he remarries or rejoins the wife or fiancée who was left behind (W). This pattern contextualizes many hundreds of Return Songs in the Yugoslav tradition and offers a proven avenue for genre-dependent comparison of a sensitive, exacting kind between Serbo-Croatian and other literatures.[29]
The principle of tradition-dependence can also open pathways to a better understanding of the poetics of Ropstvo no. 4. A rigorous philological scrutiny of the text will, for example, yield information about typical prosodic, phonological, and dialectal qualities; preparing a trial edition of a sung text like Ropstvo no. 4 is one way to underline those properties. Often something so apparently trivial as the group of sounds that the guslar employs as hiatus bridges or the phonological dynamics of an error in inflection can lead to a larger apprehension of the poem's context. Once this scrutiny has been completed, comparisons with texts from other traditions, admitted as comparands in accordance with the other steps in the reading program, can be productively undertaken. As the second principle associated with tradition-dependence, the crucial point is to give each tradition its due: for instance, one cannot speak simply of some archetypal construct called "the formula"; rather, one must describe each verbal pattern in terms of the prosody in symbiosis with which it is made and remade. Comparatists wishing to evaluate the similarities between formulaic structure in Ropstvo no. 4 and an Old English poem, for example, must acknowledge the differences between the epic deseterac and the Anglo-Saxon alliterative line (see chapter 3). Clomplementarily, the nature of the theme in Ropstvo no. 4 and of that in, for instance, the Odyssey or the Iliad must be studied in the same tradition-dependent manner. Finally, even a comparison at the most abstract level of story-pattern must take note of differences within traditions in order to make a valuable contribution to context. The remaining aspect of tradition-dependence is that of the relationship
[29] See, e.g., Coote's (1981) comparative analysis of Serbo-Croatian, Russian, and Turkish songs of Return.
of a text to its more immediate context, the singer's repertoire and the local tradition, both of which were discussed above.
In applying the last of the principles in our reading program, that of ascertaining both synchronic and diachronic contexts, I would cite first the "historical" demonstration of a cognate for the Return Song in ancient Greek and the deduction that the pattern is thus of Indo-European origin.[30] In the case of the Ropstvo , clearly a member of the same epic subgenre, then, we must posit a long pre-textual history to complement any synchronic "dialectology" assembled by reference to the recently recorded tradition. Another diachronically based observation would have to do with the relationship between the Serbo-Croatian deseterac and a reconstructed Indo-European verse form.[31] Scholars seem to agree that the epic decasyllable is extraordinarily archaic in certain basic features, apparently preserving many Indo-European characteristics in nearly Ur-form. This aspect of context presents an opportunity for a comparative evolutionary commentary to correspond with a synchronic portrait of formal metrical features in various epic lines.
2. A second type of text, the ancient Greek Return Song, has only one surviving example, the Homeric Odyssey . Here the question of text is not as simple as in Ropstvo no. 4, for, as we shall see in chapter 2, the manuscript history of the Odyssey is very much a piecemeal record. This uncertainty about the text extends to the second principle of our reading program: a final judgment that this text is taken directly from oral tradition is not defensible on the basis of the present evidence, and so we must settle for the category of "oral-derived" if the integrity of the reading process is to be observed. As I noted above, however, this does not by any means rule out the application of oral traditional theory.[32] We can and indeed must apply the methods learned from analysis of other oral and oral-derived texts in an effort to recover the traditional character of the Odyssey , that aspect of the great poem which depends for its active appreciation on an understanding of the poetic tradition. Because the Odyssey is (at least) an oral-derived text, we need to discover the nature of its traditional forms, whether they be formulas, themes, story-patterns, or whatever. In myriad ways, the traditional forms that make up the text condition and even generate meaning by establishing a background or context for the nominal, situation-specific actions and realities of the story.
The criterion of genre-dependence limits the potential comparands for the Odyssey to epic songs, and in particular epic songs of Return, for the ancient
[30] This is Lord's hypothesis (1969, 19), confirmed by evidence presented by others (e.g., Coote 1981).
[31] See esp. Jakobson 1952; also chapter 3 below.
[32] Nor, in fact, does it rule out the possibility that the Odyssey text is truly oral traditional. In setting the calibration of our reading program, however, we should make only the minimal, ascertainable assumption, thus avoiding an unnecessary and finally unprovable assertion that would weaken the whole process of interpretation.
Greek work also follows the five-part pattern of absence, devastation, return, retribution, and wedding located first in Ropstvo no. 4. Within the Homeric tradition the only other fully preserved epic is the Iliad , but it is of course quite differently structured from the Odyssey . The epic tale of the Trojan adventure has its own story-pattern, a Withdrawal-Devastation-Return sequence, and thus operates on different principles at the level of overall narrative movement.[33] This means in turn that the thematic contents of the Iliad and Odyssey must vary correspondingly.[34] Formulaic inventories of the two poems will compare more closely, since traditional patterning at the level of the line is not as attached to story-pattern as are larger units, although some scholars feel that poetic diction, in addition to being partially determined by its function in a given storytelling situation, is also conditioned by the dramatis personae themselves.[35] And philologists have long been aware of linguistic differences between the two epics (see Janko 1982). What genre-dependence prescribes, then, is a comparison of the Odyssey and the Iliad that is carefully weighted to include the discrepancies cited, and also a cross-traditional comparison with Return Songs from another Balkan area.
Our fourth principle, tradition-dependence, helps to insure the validity of the cross-traditional process by leavening the obvious similarities between works with a respect for tradition-dependent characteristics. As remarked above, we shall need in such cases to take a hard look at all comparands in the original languages from an exacting philological perspective. Many of the other problems attendant on consideration of Serbo-Croatian epic are, however, not applicable to the Odyssey : we have no firm knowledge of national, local, or idiolectal traditions with which to rationalize the dialect mixture, simply because of the paucity of surviving texts. Accordingly, the synchronic context is limited to the 28,000-line corpus of Homeric epic, with the qualifications already discussed. Diachronically, we may investigate many of the same structures mentioned before in reference to Ropstvo no. 4. The evolutionary identity of the Homeric hexameter and its historical relationship to phraseology, for example, have been subjects of fruitful debate (see further chapter 3). The field of comparative Indo-European mythology, at the other end of the spectrum, may eventually offer another aspect of diachronic context.
3. In moving from the Homeric Odyssey to the relatively minor poetic genre of the magical charm or spell in Serbo-Croatian and Old English, I take
[33] On this pattern see Lord 1960, 186-97; Nagler 1974, 131-66.
[34] For excellent studies of typical scenes and other structures in the Iliad and Odyssey , see Fenik 1968, 1974.
[35] Thus the long battle over the language of Achilles and the question of whether there is in general the possibility for "literary" manipulation of phraseology to reflect specific characters. See the opening salvo by Adam Parry (1956); contributions to the ensuing discussion include Donlan 1971; Reeve 1973; Claus 1975; Hogan 1976; Friedrich and Redfield 1978, 1981; Duban 1981; Messing 1981; Nimis 1986.
advantage of a chance to demonstrate the range and sensitivity of the reading program. For here the question of text conditions the proceedings in very definite ways from the start. The Old English magical charms, to take the more straightforward case first, are found in various medieval "leechdom" manuscripts of uncertain provenance,[36] and their structure and content are unusual enough to cause a great variety of textual problems for prospective editors and commentators. In addition to what may seem an erratic prosody, scholars have also to deal with charm texts whose diachronic identity stems from an apparent overlay of Christian elements on an earlier Germanic base. Though much of the overlay characteristically takes a prose form, the layers are not always separable, and one can be uncertain of the proper context for explication. The Serbo-Croatian spells, in contrast, generically and collectively known as bajanje by those who practice and make use of them, have been sparingly collected by native investigators and then published only in fragments. In 1975, however, a research team of which I was a member made a reasonably extensive recording of magical charms in and around the Serbian village of Orašac in the region of Šumadija south of Belgrade.[37] This collection of texts, which includes many alternate versions of the same spell by the same and different informants, provides an opportunity to study multiformity in a non-narrative genre. In comparing these Yugoslav texts with those drawn from the Anglo-Saxon leechdoms, we must be careful to allow for the inherent variance between recorded oral texts (complete with the expected "blemishes" in syntax and morphology typical of unedited material) and those edited from manuscripts of uncertain provenance.
These observations lead directly to the question of oral versus oral-derived. Only the Yugoslav charms are known oral material; the Anglo-Saxon spells must be placed in the oral-derived category, and the comparison must admit the possible discrepancies. As far as genre-dependence is concerned, we may feel confident that the comparison and contextualization are as exact and fair as possible in the present state of knowledge, especially since certain charms, like the Old English "Nine Herbs" remedy and a Serbo-Croatian spell against the incursion of nine windborne diseases, show marked similarities beyond the formal equivalence of genre.
Tradition-dependence manifests itself most immediately in the prosody that informs each body of material. The Yugoslav poems tend toward a symmetrical octosyllable, as do the lyric or "women's" songs (zenske pjesme ) also performed
[36] For a convenient edition, see Grendon 1909.
[37] This research was undertaken by an interdisciplinary team of anthropologists and specialists in oral tradition (Robert P. Creed, Barbara Kerewsky Halpern, Joel M. Halpern, and myself) with the support of a grant from the National Endowment for the Humanities. For accounts of the fieldwork and analyses to date, see Kerewsky Halpern and Foley 1978; Foley 1977a, 1980b, 1981c, 1982.
almost exclusively by women,[38] and they demonstrate a bewildering array of sound-patterns, including full- and near-rhyme, assonance, consonance, and phonetic series, as well as formulaic and syntactic frames. The Old English magic tends in its verse component toward the standard alliterative line of Old English poetry, but in this case as well a great many sound-patterns stretch the charm line out of its conventional metric: single half-lines, hypermetric verses, and seemingly faulty units of all kinds are the result.[39] The last step in our reading program, that of formulating a diachronic description to complement synchronic analysis, awaits fulfillment.
4. With the Old English epic Beowulf we enter a complex and much-discussed area. The question of text is apparently simple enough: the poem exists uniquely in the Cotton Vitellius A. xv. manuscript. But we do not know how that probably tenth-century copy (most scholars consider it a copy of an earlier text) came to be—either how the poem was originally created and recorded or how it might have been edited or recopied. Because of this lacuna in our knowledge of manuscript history (to be considered in detail in the next chapter), we cannot confidently and justly claim certain orality for Beowulf . Nevertheless, more than enough work has been done since Francis P. Magoun's pioneering article of 1953 to certify that we are dealing with at least an oral-derived text.[40] Traditional features of formula, theme, and so forth abound, and the heroic magnitude of the epic depends intimately upon their resonance.
To be sure, Beowulf meets most criteria for the epic genre as commonly cited in comparative literature studies. But, this much said, we must look long and carefully for real analogs. Some of the Old Norse material has a closely related mythic content, as do poems from other Germanic language traditions, notably the Middle High German Nibelungenlied and the Old High German Hildebrandslied ; but these works do not match Beowulf in genre. Within the Old English poetic canon only the fragmentary Waldere and perhaps a verse hagiography like Andreas can provide possible comparands, but there arc serious problems with text and genre, respectively, even in these instances. Epics in other, unrelated poetic traditions, where we find a match in story pattern and thus in epic subspecies, may hold out another alternative. Whatever the case, it is important to remember the generic singularity of Beowulf when comparing it to other works in the same tradition or in closely related Germanic traditions.
Tradition-dependent characteristics abound in the Old English epic. Its prosody and narrative structure bear some resemblance to counterparts in
[38] As compared to the epic songs (epske pjesme ), usually performed by men only and in the decasyllabic meter (epski deseterac ). For exceptions to this differentiation by sex, see Murko 1951, 189-205. For more on the epic meter, see chapter 3 below.
[39] For a discussion of these phenomena, see Foley 1980b.
[40] For a history of formulaic and thematic studies on Old English poetry, see Foley 1981b, 60-91 107-22; Olsen 1986, 1988.
ancient Greek and Serbo-Croatian, particularly in the recurrence of core ideas that underlie repetitive phrasing and narrative structures—but even that resemblance has been much overestimated. As we shall see in later chapters, the "formula" in Beowulf is much more a process than a fact, much more a continuum that extends from the level of favored metrical patterns up through their reflection in phraseology than a system of diction alone. From a synchronic point of view, the recurrent kernel of Beowulfian diction will seem to be the root morpheme, a still point surrounded by a looser aggregation of relatively less important words. The cores of themes, then, will often be key words and short phrases of great flexibility, all attracted by an abstract idea-pattern, without the kind of line-for-line verbal correspondence among instances that is typical of many themes in the Serbo-Croatian epic tradition. It is crucial to any aesthetically sound interpretation of Beowulf that we recognize these traditional features in their idiosyncratic Old English form, and that we dispense with the often undertaken but pointless search for what amounts to a tradition-dependent form from another tradition (for example, the Homeric Greek formula, the canonical unit). Once recognized, these traditional features can provide the most immediate context for a successful reading of the poem.
As is often the case in dealing with an Old English poem, the diachronic perspective must focus on a Germanic structure overlain by, and often blended with, a Christian sensibility. In Beowulf this is a particularly knotty problem, since in this poem the two are so masterfully integrated. Nonetheless, comparative diachronic analysis of the kind suggested by Alain Renoir (esp. 1981b, 1986) can create a traditional context to assist in a faithful reading of a work, particularly in those cases in which, as in Beowulf , the conventional contextualizing instruments of literary history are virtually useless because of lack of information. Finally, I would mention the research on Indo-European meter and at least pose the question of the relationship between alliterative verse and a possible ancient precursor (see further chapter 3). We can be sure of one measurement along these lines: the alliterative meter has evolved much farther away from the reconstructed Indo-European syllabic and quantitative line than have corresponding verse forms in ancient Greek and Yugoslav epic, an evolutionary history traceable to the proto-Germanic shift of stress and consequent redefinition of the metrical "prosodeme" from syllable to initial stress. A great deal more work remains to be done in this last area.
5. The final exemplar text to be considered in the light of the reading program for traditional texts is the Old English Seafarer . The unique copy of the poem forms part of the Exeter Book manuscript, a miscellany of various kinds of works ranging from folk riddles through short lyric poems to the macaronic Latin and Anglo-Saxon Phoenix and variations on the antiphons for Advent in the Roman Breviary. We know that the manuscript as it presently stands is a tenth-century collection donated by Bishop Leofric to Exeter
Cathedral, but we know nothing of the history of the making of The Seafarer previous to its inclusion in the codex. Both because of that uncertainty and because of the ample evidence in the poem of the poet's Latin learning, we must ascribe it to the oral-derived category of texts. In fact, the appearance of traditional and non-traditional structures side by side in the same text has led not a few scholars to accuse The Seafarer of inconsistency or mixed modes. But as I have argued elsewhere, the presence of two kinds of forms for articulating meaning need not prompt an accusation of interference between the two.[41] Rather, we may understand the traditional and non-traditional as simply two different and complementary ways of evoking meaning, the one through reference to the poetic tradition and the other through reference to religious tradition and the deployment of rhetorical figures borrowed from the Latin ars rhetorica .[42] The text is, in modern critical terms, "active"—that is, it calls for a complex, multi-leveled response on the part of the reader, and it offers an aesthetic experience which takes shape along two distinct phenomenological axes.
The hybrid nature of the diction extends to the poem's genre as well. The Seafarer has been termed an elegy, an allegory, an ascetic journey of expiation (a peregrinatio ), and a medieval planctus . But we would be most accurate to view each of these labels as apposite rather than as all-inclusive, and to understand that The Seafarer is a work sui generis . Its truest comparands in Old English are poems which answer some of the same generic criteria and which demonstrate a similarly hybrid phraseology, rhetorical structure, and Christian sophistication; among them would be numbered The Wanderer, The Wife's Lament , and The Husband's Message . Outside the Anglo-Saxon tradition, the shorter Christian poems from Serbo-Croatian oral tradition, which also show traditional forms in the hands of a gifted artist, offer a promising analog (see Foley 1983b). Finally, many of the same tradition-dependent and diachronic considerations may be applied to The Seafarer as were applied to Beowulf Again the pagan-Christian blend is most prominent here, and is of course intimately connected with other parts of the reading program.
From this point on I shall turn away from both general issues in contemporary oral literature research and the reading program for traditional texts in order to conduct a more focused examination of epic poems from the ancient Greek, Old English, and Serbo-Croatian traditions. In accordance with the reading program, we shall first look at the comparability of the documents themselves in order to assess as exactly as possible what authority each has in presenting its oral or oral-derived poem.
[41] See Foley 1983a,b. The co-existence of oral traditional structure and Latin learning in the same text may be explained in part by the cultural diglossia prevalent in the medieval period; see further Stock 1983 and Ong 1986.
[42] For examples of the latter technique, see Campbell 1966, 1967, 1978.