Nine
Thematic Structure in Beowulf and Old English Poetry

In concert with the method used in earlier chapters, I propose to examine the Old English narrative theme as a tradition-dependent entity, that is, as a compositional unit that depends for its structure not only on the general oral-derived nature of Old English poetry but also and equally on the idiosyncratic features of the Anglo-Saxon language and the specialized formulaic diction that serves as its expressive medium. Toward that end we shall first review the scholarship on the Old English theme before turning to analysis of sample passages from Beowulf and another long narrative poem from the canon in order to determine precisely how we should view and define the unit. The themes to be analyzed will include the sea voyage and the scourging of the hero. In considering the last of these, drawn from the verse hagiography Andreas , we are reaching beyond the poem Beowulf , and indeed beyond its genre of heroic epic, to the ubiquitous saint's life. But in doing so we are at least turning to a long narrative poem that may safely be considered oral-derived,^[1] thus maintaining a reasonable degree of congruency in the comparison. Were the surviving Anglo-Saxon poetic canon larger, the comparison could be made more exact.

Prior Research

Origins

As with so many other aspects of research on oral traditional structure, scholarship on the Old English theme began with Albert Lord's 1949 disser-

[1] On the traditional nature of the phraseology, see, e.g., Peters 1951; note also the poet's interruption of his narrative at 1478-91, in which he discusses the compositional problem of making his poem lytlum sticcum ("in little pieces"). The major part of this chapter ("Two Example Themes") will provide additional evidence of the oral-derived nature of the poem at the thematic level.

tation, eleven years later to become The Singer of Tales .^[2] Defining the general, cross-traditional unit as a "group of ideas regularly used in telling a tale in the formulaic style of traditional song" (1960, 68), he went on to illustrate the dynamics of the theme in Serbo-Croatian, ancient Greek, Byzantine Greek, Old French, and Old English. Moreover, because the published version of his account appeared some five years after thematic analysis in Old English poetry had formally been begun by Francis P. Magoun and Stanley B. Greenfield, Lord was able to distinguish his concept of the unit from theirs. With typical care and precision he remarks (pp. 198-99):

I should prefer to designate as motifs what they call themes and to reserve the term theme for a structural unit that has a semantic essence but can never be divorced from its form, even if its form be constantly variable and multiform. It is not difficult to see that even from this point of view there are themes in Beowulf : repeated assemblies with speeches, repetition of journeying from one place to another, and on the larger canvas the repeated multiform scenes of the slaying of monsters.

He goes on to give additional examples, such as the arrival of the hero, which along with the assembly is also common in Serbo-Croatian epic, and in general to limit the conception of the theme to an integral narrative unit whose beginning, middle, and end are well defined.

In the meantime, between the submission of Lord's dissertation and its publication, Magoun had isolated what he took to be an oral-formulaic theme: "the mention of the wolf, eagle, and/or raven as beasts attendant on a scene of carnage" (1955, 83).^[3] Finding twelve occurrences of this multiform in the canon, he then subjected them to a phraseological analysis, concluding (p. 90) that "the formulas and formulaic systems will be seen to divide up in two ways, those potentially relevant to the subject matter of the theme and those of general usefulness." While this concatenation of narrative details may represent a theme for Magoun, it is clear why Lord would rather consider it a motif: there is no repeated, specific set of actions involved in the "Beasts of Battle," no narrative process preserved from one instance to the next. This is, by comparison to "assembly" or "readying a hero's horse" in the Serbo-Croatian epic tradition, a somewhat static collection of details uncatalyzed by a network of recurrent and associated actions.

Greenfield's (1955) theme of "Exile,"^[4] more like Magoun's than Lord's

[2] For a fuller account of the background scholarship than is possible here, see Foley 1981b, 1985, 1988; Olsen 1986, 1988. A discussion of Lord's specific proposals may be found above in chapter 8.

[3] It is well to recall that in 1951 Magoun (p. 84) had pronounced Beowulf and other Old English poetry to be certainly oral, by virtue of the dictum that "oral poetry, it may be safely said, is composed entirely of formulas, while lettered poetry is never formulaic."

[4] See also his two earlier articles (1953a,b), which adumbrate thematic analysis without associating the structure with oral tradition.

unit, consists of the four elements status, deprivation, state of mind, and movement in or into exile. Again we encounter a collection of details without a clear narrative matrix, a recurrence of what Greenfield calls exile "images" without a specific network of actions. And once more it is apparent why Lord would choose to call such a unit a "motif" rather than a theme.

The question that these early studies presage is essentially the burden of the present chapter: are all Old English themes really oral traditional units, and how does the principle of tradition-dependence affect the definition? While typologies constructed for their own sake (too often against the empirical evidence) are certainly of limited usefulness, it is equally misleading to dissolve all distinctions in search of an overarching generalization. What we shall seek in the analytical section of this chapter is a workable, tradition-dependent idea of the Old English theme, one that recognizes both the distinctions made by Lord and the discoveries reported by Magoun, Greenfield, and others—for the lesson of the earliest scholarship in this area is that we must be willing to undertake a true comparison, one that can draw on findings in other oral traditional literatures but must nevertheless be rigorously grounded in the philology of its own tradition.

A Brief Summary of Scholarship

Building on the foundations erected by Lord, Magoun, and Greenfield, later scholars have uncovered numerous themes in the narrative poetry of the Old English canon. In lieu of extended discussions of these contributions, I present below a digest of the twenty-four themes or type-scenes reported to date, together with references to the articles devoted to each. A general overview of the most frequent or most often cited themes follows.

Far the best-documented thematic unit (fifteen citations), the "Hero on the Beach" was firs reported by David Crowne in 1960 and has since been shown m recur not only throughout the Anglo-Saxon poetic corpus but also in Middle English and Old High German.^[5] Crowne (p. 368) defined this compositional pattern as "a stereotyped way of describing (1) a hero on the beach (2) with his retainers (3) in the presence of a flashing light (4) as a journey is completed (or begun)." This account of the content approaches Lot's concept of the theme more closely than does Magoun's "Beasts of Battle" or Greenfield's "Exile," since the repeated action is more clearly delineate. But if narrative sequence is part of Crowne's idea of the unit, verbal correspondence among instances most certainly is not. In fact, he goes m some length to demonstrate that the theme in Homer as well as in Old English amounts to a grouping of ideas rather than a critical mass of phraseological items, concluding (p. 364) that the theme "does not depend upon a fixed content of specific formulas for its mnemonic usefulness m the singer." This concentration on narrative fabric as the stuff of which the compositional theme is made—and the abandoning of verbal correspondence among instances as a criterion for identification of the unit—marked a point of departure for studies of the theme or typical scene in Old English poetry. No longer did investigators search for repeated verses as the telltale sign of the recurrent scene; in practical terms, they assumed an unlimited variety of situation-specific instance, with a variety of diction m match. The theme in Old English became purely a sequence of ideas for those following Crowne's lead, and scholars were virtually unanimous in doing so.

Magoun's "Beasts of Battle" has also proved important in the development of thematics in Anglo-Saxon verse. Assuming that the Beowulf poet was lettered but composed formulaically, Adrien Bonjour contended in 1957 that this poet,

[5] For a complete survey of oral-formulaic scholarship in Middle English, see Parks 1986. For Old High German, see Renoir's analysis of the Hildebrandslied (e.g., 1979b).

unlike his lesser contemporaries, shows artistic originality in the handling of the "Beasts" pattern. This general notion of the theme as a structural entity that could be shaped according to individual aesthetic design found apparent support in, for example, Robert P. Creed's 1961 article on oral poetics, in which he argued that a given theme should be understood not simply in relation to other instances within a given poem, but also against the larger traditional background.^[6] Thus arose the concept of aesthetic manipulation of traditional themes, just as it had for the formula some years earlier. As time went on, specialists in Old English, it is fair to say, were quite willing to accept a narrative unit such as the theme as long as the theory allowed for the poet's conscious artistic control of his traditional medium.

A third significant step in the study of Old English themes is that taken by Alain Renoir in his contextual analyses of various medieval texts.^[7] Although one cannot tie his method uniquely to one particular multiform, his efforts have concentrated on elucidating the expressive content (and this means audience expectation as well as textual structure) of the "Hero on the Beach," "Beasts of Battle," and "Speaking Wood" patterns. Renoir's method entails establishing an inter-textual and often cross-traditional directory for a given oral-formulaic theme, even if it occurs in a written text (e.g., 1976b), and interpreting each instance in the context of what is collectively known about the thematic pattern. In the case of poems whose authorship or provenance is uncertain—and this is often the case with medieval works—this approach provides otherwise unavailable insights into a great range of critical issues as various as manuscript authority and rhetorical structure.

Current Issues

At the moment the most pressing issues in Old English thematics are (1) the distinction proposed by Donald Fry between "theme" and "type-scene" and its implications, (2) the role of verbal correspondence among instances in the definition of the unit, and (3) more generally, the question of whether we really have a "theme" in Old English verse—and if so, how it is to be reconciled with the obviously quite different comparands in ancient Greek and Serbo-Croatian. While I shall depend on the analyses below to furnish answers for these questions, let me first present each issue in sharper focus.

In 1968 Fry proposed a typology of narrative units since advocated by many scholars. He differentiated (p. 53) between the type-scene , or "recurring stereo-typed presentation of conventional details used to describe a certain narrative

[6] With reference to the Serbo-Croatian analog as discussed in chapter 8 of this volume, we may note that the guslar possesses an idiosyncratic, idiolectal form for themes that are in general use throughout his local tradition and in the poetic tradition at large.

[7] See, e.g., Renoir 1976a, 1976b, 1977, 1979a, 1981b, and esp. 1988.

event, requiring neither verbatim repetition nor a specific formula content," and the theme , or "recurring concatenation of details and ideas, not restricted to a specific event, verbatim repetition, or certain formulas, which forms an underlying structure for an action or description." While both derive from Crowne's (1960) conception of a unit undefined by verbal correspondence among instances, Fry's two units seem to diverge sharply, and their divergence is precisely the presence or absence of a narrative context. If the repeated collection of details is imbedded in an action-pattern, it is an example of Fry's type-scene; if it lacks such a narrative matrix, it is his theme. We may recall the distinction made by Lord at this point and note that Greenfield's "Exile" (and perhaps Magoun's "Beasts of Battle" as well) would be classed by Fry as themes, while Lord's examples of Assembly and so forth would be termed type-scenes. This distinction thus made, one may ask how useful it proves in analyzing and interpreting Old English poetry.^[8]

Our second issue concerns the phraseological aspect, if any, of the Old English theme. In opposition to the preponderance of critical opinion, which following Crowne holds that themes in Anglo-Saxon verse are narrative sequences without any particular formulaic content, Lord has argued that a true theme must exhibit not only a characteristic grouping of ideas but also a definite verbal consistency. It is primarily on these grounds that he distinguishes between the literary concept of theme and the oral traditional unit (1975, 20):

If, however, by theme one means subject , a narrative element, such as a catalogue, or a message, or equipping, or gathering an army, then our definition is inadequate; for clearly we can find gatherings of armies, equippings, messages, and catalogues in written as well as in oral literature. The theme as subject alone is too general for our very special purposes. But if by theme one means a repeated narrative element together with its verbal expression, that portion of a poem, an aggregate of specific verses, that tells a certain repeated part of the narrative, measureable in terms of lines and even words and word combinations, then we find ourselves dealing with elements of truly oral traditional narrative style.

As we enter the analytical section of this chapter, this issue of verbal correspondence will continue to be a concern: namely, do we demand an aspect of verbal recurrence for the theme and, if so, what do we make of the patterns that show no such verbal recurrence?

This question leads directly to the last of the issues to be kept in mind as we examine narrative patterns from Beowulf and the Old English poetic canon. Whatever conclusions are reached about what an Old English theme is or is not, we shall have to confront a real discrepancy between the structure and texture of narrative units in Old English on the one hand and in ancient Greek and Serbo-Croatian on the other. That is, even if we were to so tighten

[8] As illustration, see Fry 1969.

the definition of theme in Anglo-Saxon verse that it admitted only a small number of the twenty-four patterns summarized above, we would still be left with a question more important than the matter of definition. In short, just what function do these other, "non-thematic" units perform, and more specifically, what is their role in Beowulf and other poems? Are they traditional and (at least) oral-derived, or do they represent some sort of literate extrapolation of bona fide oral traditional units? These questions have clear implications for the aesthetic study of the poems as well.

The Tradition-Dependent Nature of Old English Themes

From a practical point of view, the concept and definition of the theme as developed by Albert Lord in respect to Serbo-Croatian oral epic translated more easily to ancient Greek than to Old English oral-derived poetry. To appreciate why, we need only remember the two aspects of the compositional unit so carefully defined by Lord in 1974: narrative sequence and verbal correspondence. While a sequence of narrative ideas may, unlike phraseology, be free from the idiosyncratic shaping exerted by language and prosody, so that an action-pattern should be able to be defined in approximately the same terms whatever the language of the given tradition, the necessarily varying phraseologies that express those action-patterns will not submit to a single universal concept or definition. And as we have seen from earlier chapters, the phraseological leap from Serbo-Croatian to Homeric Greek is much shorter than from either of those epic poetries to Old English. Boewulf and other Anglo-Saxon poems reveal a formulaic diction, to be sure, but it does not follow the prosodic rules observed by Homer or the Yugoslav guslar . If this tradition-dependent phraseology serves as the idiom for the actual verbal expression of thematic units (as opposed to their identity as groups of ideas only), then as a first approximation we should expect that verbal correspondence among occurrences of the Old English narrative unit will prove similarly tradition-dependent.

Along with this predictable difference, it is well to recall the nature of multiformity revealed in our close examination of the Serbo-Croatian epic theme in chapter 8. Instead of simple overlays of units, we found that the Yugoslav tradition can be faithfully described only by distinguishing among idiolectal, dialectal, and tradition-wide versions of a given theme. In other words, the theme exists as a replicating unit—one that recurs preserving both the overall grouping of ideas and a significant amount of diction associated with those ideas—most obviously within the repertoire of the individual singer, who has his own "version" of the typical scene. More variation is observed at what has been termed the level of "dialect," that version of a theme employed not by the individual but by his local traditional community, the

singers in his geographically defined district. Still greater variation characterizes the narrative multiform at the tradition-wide level, as one compares versions of a theme from singers in different, extensively separated regions. It must be emphasized that such distinctions, possible in such a well-recorded epic tradition as the Serbo-Croatian, cannot be recovered in dead-language, manuscript-based poetries like ancient Greek and Old English.^[9] Our uncertainty as to the "authorship" of various poems in the Old English canon, not to mention our probably irresolvable doubt about the actual role of writing in their composition and transmission, limits the extent to which we can compare structural details on a one-to-one basis. It will be most prudent first to attempt a careful characterization of the Old English theme as a tradition-dependent entity, realizing that a philologically sound definition may vary considerably from those formulated for ancient Greek and Serbo-Croatian, and then to collate those observations with earlier discoveries.

One more caveat is in order before beginning analysis. As earlier chapters have attempted to show, the assumption that all parts of even a verifiably oral epic are uniformly formulaic is untenable. Likewise, even the Serbo-Croatian theme (as observed in the work of a single singer and in a single community of singers) does not prove uniformly dense in verbal correspondence; certain parts of the unit are, for one reason or another, repeated more exactly than others from one occurrence to the next. These variations are traceable to a number of sources, among them the singer's deployment of sound patterns and his familiarity with a given theme. The point here is that an interactive, purely synchronic model for formulaic diction and for narrative multiformity in Serbo-Croatian oversimplifies the real situation. And if we should be prepared for variation in formulaic density and degree of verbal correspondence in this ascertainably oral tradition with its treasury of recorded performances, then we should also anticipate similar fluctuations and variations in the manuscript texts of finally uncertain provenance.

Two Example Themss

In order to determine the tradition-dependent shape of the thematic unit in Beowulf and Old English poetry, let us consider two narrative multiforms, the "Sea Voyage" and the "Scourging" scene, the first from Beowulf and the second from the verse hagiography Andreas . In each case the focus will be on measuring the importance of idea-structure on the one hand and of verbal correspondence on the other.

The Sea Voyage in Beowulf

In examining this multiform in Beowulf , it is well to recall four basic properties of the Serbo-Croatian epic theme as a comparative background. First,

[9] On thin point cf. Gunn 1971.

as would be expected given the colon structure underlying formulaic phraseology in this epic tradition, any verbal agreement among instances of a narrative multiform is expressed largely in terms of whole lines or cola, as in Homeric epic. The entire line and caesura-bound hemistich are thus the "lexical" data of the theme. Second, although narrative sequence seems in all cases to remain almost absolutely constant and to oversee the thematic progress of the story, in rare cases a unit can be transposed, provided that the narrative logic is maintained. Third, verbal correspondence is not of uniform density throughout the theme or from instance to instance; certain motifs are much more stable formulaically than others. Fourth, variation in verbalization of the theme can take a number of forms, among them what may be termed "formulaic variance," in which lines and parts of lines recombine according to systemic principles, and "ornamentation," in which a non-narrative, paratactic gloss not necessary in itself but rather complementary to a necessary element is included.^[10]

The Old English theme is equally dynamic, though in its own tradition-dependent way. Its texture is also uneven, with certain sections exhibiting one level of verbal correspondence and others differing. Although we do not have the luxury of multiple texts provided by the well-collected Serbo-Croatian tradition, the typical scene of the Sea Voyage will help to make this point. The two occurrences of the sea voyage in Beowulf may be rationalized by the motif structure given below.^[11]

[10] See chapters 5 and 8. On other sorts of patterns, such as phonemic series, end-colon rhyme, and semantic association, which also characterize other oral genres in the Serbo-Croatian tradition, see, e.g., Kerewsky Halpern and Foley 1978.

[11] On the texts of the following passages, I follow Klaeber (1950) except for his proposed emendation to naca in line 1903b (nom. sing. and subject), for which I see no clear justification. I choose to return to the manuscript reading nacan , ostensibly a dative singular taking its inflection from the preceding on .

[12] The batweard ("boat-guard") I take to be the same person as the landweard , ("land-guard," 1890b), since it is the landweard who promises to guard Beowulf's ship until he returns (see Fines 293-98).

The basic narrative structure of the Sea Voyage may thus be represented as a sequence of five elements:

A. Beowulf leads his men to the ship
B. The ship waits, moored
C. His men board the ship, carrying treasure
D. Departure, voyage on the sea, arrival
E. They moor the ship

Within this regular structure are interwoven elements that help to particularize the generic action, to suit the theme to its narrative context. The W motif, "Armor," variable in position, seems to serve as the prerequisite for the X motif, "The coast-guard approaches"; where the latter first occurs, the former closely precedes it. In occurrence #1 above, W introduces X within the E element, and in #2 W is followed by X before the G element. In narrative terms, this means that "Armor" and its consequent can be developed during the mooring of the ship (as in #1) or before Beowulf's men carry the treasure on board (as in #2), in concert with the demands of the situation.

The X motif, most protean of any of the elements, deserves close attention. It concerns an encounter with the guard of whatever coast one lands on or is in the process of leaving, and is applied to at least two distinct characters and situations in Beowulf .^[14] In occurrence #1 the story line calls for an expansion of X into a lengthy exchange between the coast-guard and Beowulf. Working within the traditional idiom, the poet thus accomplishes two things concurrently: he both fulfills the generic thematic conditions and suits the

[13] Although the Geatish hyðweard certainly not the same per, on as the Scylding figure involved in X₁ and X₂ , he performs the same generic function and can thus be seen as the agent of the thematic variant.

[14] See notes 12 and 13.

structure to its narrative environment. Beowulf is forced to identify both his people and his purpose, points of information that arc of course vital to what follows, and the identification proceeds naturally (that is, traditionally) out of the Sea Voyage multiform. This same potential for modification or variation shows up in a different way in occurrence #2, where X occurs three times (X₁ , X₂ , X*; the symbol X* is used to denote a reversal of the more usual action of this motif). In all other instances, the coast-guard approaches the Geats; in occurrence #1 he comes to challenge their disembarkation, and in occurrence #2 (X₁ -X₂ ) he again moves to greet them, though this time in a friendly manner, as the litotes (1982-95) indicates. The harbor-guard of X*, in contrast, has been watching for Beowulf and his men for a long time. He has remained on the shore, waiting for the returning heroes to approach him. While this constitutes a reversal of the encounter, it springs from the same traditional form, the X motif, as do the other cited instances.

The major motifs within this theme, that is, elements A-E, also reveal a certain amount of adaptation to context, although they vary in structural stability. Element A, for example, splits in occurrence #2, enclosing a number of details within a narrative capsule.^[15] One dimension of this division is clear: A₁ mentions nothing about Beowulf's companions, while in occurrence #1 this motif includes both the hero and his men. A₂ thus fills out the element by describing the embarkation of Beowulf's men. Motifs B, C, and D arc quite stable within the Sea Voyage sequence, especially the last of them, which is tightly organized around a three-part series of departure, voyage, and arrival. Two specific features further structure the action of the D motif: (1) the "go until one sees the destination" commonplace^[16] and (2) the notation of wind and the ship's sail. The splitting of E in occurrence #1, like that of A in #2, presents evidence of the pliability of traditional elements, as once again the recurrent generic structure is adapted to the narrative situation.

Taken together, the two instances of the Sea Voyage in Beowulf collectively exhibit a discernible and dynamic narrative sequence , an action-pattern not dissimilar in nature to that found in Serbo-Croatian themes. In applying the first of Lord's criteria to what we have in Beowulf , we seem to have a dose fit: with the exception of the Old English poet's somewhat greater flexibility in motif development,^[17] there is considerable similarity between the Old English and Serbo-Croatian themes in terms of narrative sequence. But what of the second criterion, verbal correspondence ? Are we likely to observe the same dose fit across traditions along the axis of phraseology?

[15] Such capsules or envelopes, often understood as elements in ring composition, have themselves been the object of a number of studies. See Bartlett 1935; Hieatt 1975; Niles 1979.

[16] On this common pattern, see Irving 1968, 32-42; G. Clark 1965b, 647-48; Gruber 1974.

[17] This difference may well stem from the limited textual sample in Old English, from the greater freedom in expression that is characteristic of Old English traditional phraseology, or from both.

To answer this question we must recall what was discovered earlier (chapter 6) about the tradition-dependent character of the formula in Beowulf , for with that latter principle in mind it is possible to predict another directly related tradition-dependency—the nature of thematic data. First, we cannot expect a large proportion of classically defined whole-line or half-line formulas as verbal correspondence in Old English poetry, since such an expectation presupposes a colonic formula and, as demonstrated earlier, Old English prosody and phraseology are not colonic. Second, what we can logically expect as thematic data are highly variable half-lines that may have in common only their stressed cores. What verbal correspondence exists will thus appear to take the form of single morphs, that is, of roots of words whose systemic context is metrically (and therefore lexically and syntactically) highly variable. This does not preclude formulaic content, as we shall see below, but simply makes it more likely that single words will constitute whatever thematic resonance obtains in the actual phraseology.^[18] Following is a tabulation of the morphs and the few phrases that help to define the Sea Voyage multiform in Beowulf .

A:A₁ , A₂ —none

B:B—none

C:C—sande/sande (213a/1896a), nacan/naca (214a/1896b), stefn/-stefna (212a/1897b)

D:D—Gewat/Gewat (2177a/1903b), wæg-/weg- (217a/1907a), winde/wind(217b/ 1907b), wudu/-wudu (216b/1906b), -sið/siðes (216a/1908a), sæ-sæ- (223a/1908b), bundenne/bunden- (216b/1919a), brim-/brim- (222a/1910b), sund/sund- (223b/ 1906b), flota famiheals/fleat famigheals (218a/1909a), -clifu/clifu (222a/1911a), -næssas/næssas (223a/1912a), up/up (224b/1912b)

E₁ , E₂ :E—sædon/sa le/sale/sælde (226a/302a/1906a/1917a), sid fæpmed scip/sid fæpmd scip (302b/1917), on ancre fæst/once rbendum fæst (303a/1918a), -wudu/wudu (226a/1919a)

W:W—syrcan/-syrcan (226b/1890a)

X:X— weard/-weard/-wearde/-weard (229b/1890b/1900a/1914b)

In addition to items of verbal correspondence among instances of a theme, there exists in Beowulf and other Old English poetry a tradition-dependent, local resonance of morphs that may be called responsion .^[19] Rather than being attached to a certain narrative event or pattern and echoing traditionally against other occurrences of that same event or pattern, these words respond to proximate partners, lexical relatives usually no more than about twenty

[18] I might add that these data cannot be easily explained away as words that are most likely to occur in a given description. Lexical items at positions of metrical stress are the products of a process, and that process, which does not characteristically yield colonic phraseology, should be understood on its own tradition-dependent terms. Compare the grouping of morphs in the Joy in the Hall theme described by Opland (1976).

[19] This kind of local verbal echo has received attention as a compositional and artistic feature in Old English poetry. See, e.g., Beaty 1934; Kinegen 1977; Harming 1973; Foley 1980b.

lines away, and often much closer.^[20] Although there is no opportunity in the present discussion to do more than suggest the presence of responsion and to list the occurrences in the Sea Voyage theme (see below), I would emphasize its importance to the poetics of Beowulf . For example, many rhetorical figures attributed by some critics to direct borrowing from Latin authors can be derived from the interaction between responsion and other aspects of Germanic verse form.^[21]

[20] Of course, if a motif is split, as is the case with E₁ -E₂ in Occurrence #1, a greater distance between responsions can result.

[21] Cf. Campbell 1966, 1967, 1978. Even some of the figures that Campbell cites frequently as examples of borrowed techniques (anaphora, polysyndeton, hypozeuxis, etc.) can be so explained. The subject deserves much fuller treatment than is possible here, and I must for the moment leave the argument at the level of suggestion.

If the Beowulfian Sea Voyage and the Serbo-Croatian and ancient Greek themes resemble one another quite closely at the level of narrative sequence, they diverge considerably with respect to verbal correspondence. To be sure, there are some hard lexical data (of similarly uneven distribution) for all traditional themes, but these data take quite different, tradition-dependent forms. The Old English correspondence manifests itself chiefly in morphs in positions of metrical stress, with a less strictly defined, more variable phraseological environment. Occasionally a half-line formula—understood in the classical colonic sense—occurs, but this is relatively rare. What correspondence exists in Serbo-Croatian and Greek, in contrast, consists of lines and cola—that is, of the bound, encapsulated phrases that form in symbiosis with a syllabic and consistently demarcated metric. Neither the Sea Voyage nor either of the other two units is less a theme for its similarity to or divergence from its counterparts, rather, each theme takes shape in a form governed by the prosody of the tradition involved.^[22]

[22] of the three additional occurrences of this theme cited by Ramsey (1971, 56)—Andreas 230-53, 349-81, and Elene 212-75-I would include only the last as a true instance of the Sea Voyage. The Andreas passages contain little of the same narrative pattern examined above and none of the phraseology; perhaps because of the influence of the Praxeis source, which calls for an extended conversation among the principals during the voyage as an important part of the story to be told, the five-pan structure was not useful and did not occur. The Elene passage, in contrast, if abbreviated to its core (lines 225-55), does exhibit the characteristic narrative sequence: A—The hero(ine) leads her men to the ship (225-26a, 229-32a); B—The ship waits, moored (226b-28); C—The men board the ship, carrying treasure/armor (232b-36); D—Departure, voyage on the sea, arrival (237-50a); and E—They moor the ship (250b-55). There is enough variation from the Beowulf occurrences to suggest the influence of a source, a different poet or local tradition (as in the Serbo-Croatian examples in chapter 8), or both.

One further example of the Sea Voyage in Beowulf may help both to broaden our understanding of the unit and to point the way toward the aesthetic implications of traditional structures.^[23] Early in the poem, long before Beowulf appears, the archetypal hero Scyld Scefing embarks on a sea voyage, albeit of a significantly different sort. For Scyld the journey is his funeral, the ritual marking his passage from this world to the next. At the appointed time, we are told, Scyld "led his men to the ship" (A, 26-31), with the surviving retainers bearing him according to his previous order. In the next few lines we hear that "the ship waits, moored" (B, 32-33), and that he and his men "board the ship, carrying treasure and armor" (C, 34-48a), the hero being laid by the mast with his grave-gifts and appropriately eulogized by the poet with a litotes. Just as the C element is elaborated with descriptions of the treasure and the hero, so the D element, "Departure, voyage on the sea, arrival" (48b-50a), is correspondingly foreshortened; this is after all no customary Sea Voyage but a ship burial. Finally, instead of the expected form of element E ("They moor the ship"), we hear the agnostic profession of ignorance over Scyld's destination (50b-52):

Men ne cunnon
secgan to soðe,
seleræedend e, hæleð under heofenum, hwa pæm hlæset onfeng.

Men do not know
[how] to say in truth, hall-counselors,
heroes under the heavens, who received that burden.

Whether this particular usage of the Sea Voyage theme was a widely traditional one we shall probably never know, but the recognition of its existence at the root of Scyld's funeral enlarges our notion of thematic morphology and offers a perspective on the poet's art. First, because this narrative structure can be employed in such an (apparently) unusual way, we have more evidence of the Old English theme's consistent narrative pattern and extremely variable phraseology. Complementarily, the referential meaning of the thematic structure—Sea Voyage—imbues the funeral with a significance far beyond the actual event. Not only are the poet and his tradition conceiving of the great hero's departure as a sort of ultimate Sea Voyage,^[24] but the conspicuous absence of the traditional closure in the mooring of the ship powerfully evokes

[23] Although I happened on the thematic correspondences independently, the outward focus of the following analysis harmonizes with that of Ramsey (1971). Much of what his article presents seems sound, but I would advocate going beyond a conception of the relationships as merely habitual and convenient to understanding the referential power of the Sea Voyage multiform as an idiomatic way of describing the funeral.

[24] Cf. the extrapolation of the Storm Giant mythologem to Apocalypse in Exeter Riddle I (Foley 1976b), as well as the metonymic import of the unfaithful wife and the consequent lack of the traditional closing to the Serbo-Croatian Return Song (chapter 10).

the cosmic overtones of this last and special journey as well. This view of Scyld's passage from the world is made possible through the metonymic poetics of oral tradition, without whose associative dynamics such a perspective could not be achieved.

The Scourging Scent in Andreas

For a second example of the Old English theme I turn outside Beowulf to another long narrative poem, the apocryphal story of St. Andrew among the Mermedonians. Scholars have long recognized similarities of diction and narrative structure in these two narratives, and prevailing opinion ascribes these similarities to their shared roots in the Anglo-Saxon poetic tradition rather than having recourse to the once current literary explanation of common authorship.^[25] In extending the analysis beyond Beowulf we shall be seeking to test the concept of the theme developed in this chapter on a text in the "religious" tradition of Anglo-Saxon poetry, and therefore very likely an oral-derived rather than a primary oral text, one perhaps further removed from the native Germanic oral tradition than the less patently Christian Beowulf .

In the process we must also consider Albert Lord's judgment (1975, 23) on the Old English Christian poems and their relationship to oral composition: "If the religious poems were truly oral traditional songs, I would expect to find a higher degree of verbal correspondence among the various instances of a theme within a given poem, after making due allowance for adjustment to the specific position in the poem which it occupies." In light of what we have so far learned about the Old English theme, can we demand verbal correspondence (implicitly of the ancient Greek or Serbo-Croatian sort) as a sine qua non in this oral-derived hagiography? While there is little doubt that poems like Andreas are "transitional" or "mixed," in Lord's terms,^[26] I would urge two preliminary caveats: first, it is impossible to determine such transitional character on the basis of criteria shown to be inapplicable to Old English poetry, and second, we cannot assume without testing that even such supposedly mixed poems will not preserve the basic thematic structure we find in Beowulf , that is, the tradition-dependent thematic structure of Old English verse.

In one sense Andreas provides an ideal opportunity for such a test, since in addition to being composed in the traditional idiom it derives in some fashion from an original and surviving Greek text, the .^[27] Although critics have

[25] Compare Sarrazin 1897 with Peters 1951 and Crowne 1960.

[26] See his further remarks on "transitional texts" in Lord 1986c.

[27] Brooks (in the standard edition of the Old English poem: 1961, xv-xviii) and Schaar ([1949] 1967, 12-24, esp. 20) concur that the closest source is the Greek Praxeis ; see their accounts for descriptions of other texts of the same story. Quotations of the Praxeis are taken from Bonnet 1898.

sought to explain the divergences between the Greek version of the story and the Old English poem by invoking a lost Latin intermediary, this explanation cannot account for the larger, institutionalized (and recurrent) variations between the two. That is, while the poem's mention of Bishop Platan and Andrew's departure (both examples cited by Brooks [1961, xv]) may be the result of the poet's consultation or knowledge of other versions of the legend, the Old English poet's demonstrably traditional embroidery on the bare fabric of the Greek narrative must be understood as the contribution not of a lost manuscript but of the poetic wordhoard. As long as we stipulate, then, that divergences between the Greek and Old English texts must be institutionalized and recurrent in order to merit interpretation as traditional structures, the way is open to test Andreas for thematic patterns by superimposing the details of its story against the usually close source, the Praxeis .

For this purpose I have chosen a narrative pattern that recurs in both the Greek and the Old English texts, although, as we shall see, the actual mode of recurrency is quite different.^[28] This is the three-day scourging of St. Andrew by the Mermedonians, in both versions a thrice-repeated incident that causes the holy man much suffering as he is dragged about the city until his body is broken and bloody. The cannibalistic heathen, urged on by Satan, are attempting to break his spirit and his faith in God, but the ever-worsening punishment and torture seem only to strengthen Andrew. He survives each scourging and in the end defeats his enemies; after the third incident, God looses a torrent of water on the Mermedonians in answer to Andrew's prayer. Thus, what begins as a threat to the saint's life and faith ends in an affirmation of God's power and of human perseverance through earthly travail.

What is more, each day is a microcosm of the three-day torture of Andrew. In both versions the Scourging process begins with the onset of day and continues until sunset, at which time the still faithful saint is returned to his cell for a night of mental anguish. Whether he is assailed at night by his own doubts or by the direct attack of Satan and seven demons, the diurnal cycle marks the stages in his constant trial, which goes on for three days in basically the same fashion until the divinely retributive flood ends the scourging cycle.

Against this consistency underlying the three successive scenes in the Greek Praxeis we may profitably juxtapose the equivalent passages from the Anglo-Saxon Andreas , and the result is enlightening: in addition to the consistency attributable to the source, the Old English text reveals further, complementary recurrency of its own—a recurrency that may confidently be labeled "traditional" because it closely resembles the structure of the Old English theme

[28] The material to be analyzed is found between lines 1219 and 1462a in the Old English poem and in Chapters 25-28 of the Greek prose. See further Table 22.

as discussed above. Above and beyond the patterning that derives from the Praxeis , then, we observe a narrative patterning typical of thematic structure in Anglo-Saxon verse.

The Scourging theme in Andreas can be described as a definite series of actions, only some of which are found in the Greek source. These actions or motifs are:

A. The enemy arrives with a large troop
B. They lead the prisoner from his cell and drag him about the city
C. The prisoner's wounds are described
D. They lead the prisoner back to his cell
E. The prisoner confronts night and mental torture

The fact that these five motifs recur in each of the three Scourging scenes, regardless of what the Greek text has at the same point, indicates that this sequence had an independent existence as a narrative unit,^[29] one that could be adapted to context, to be sure, but also one that had a certain integrity as a traditional unit. Although, as already mentioned, it is only prudent to view Andreas as an oral-derived work necessarily further removed from primary oral tradition than Beowulf , with it we can begin to understand the power and longevity of oral-formulaic themes.^[30]

One example of the Scourging theme's independence is the modification it forces in the story at the very beginning of the three-day trial. In chapter 25 of the Praxeis , following God's exhortation of a few lines earlier, Andrew rises up and boldly reveals himself to the heathen multitudes: in effect, he ends their search for him by giving himself up.^[31] This brave act leads directly to the first day of torture, as a diabolically inspired cannibal suggests fastening a rope around Andrew's neck and dragging him through the streets. In the Old English poem God also instills Andrew with courage and counsels him not to hide (1208-18), but no mention is made of the saint actually delivering himself up to his enemies. Instead, what immediately follows God's encouraging and comforting speech is precisely motif A, the onset of the Scourging theme. In place of surrender to the enemy the Old English poet summons a (contextually quite inappropriately) large troop to seize the less-than-threatening saint:

[29] Gr. Lord 1960, 69-98. It may have been that the Scourging theme was used only in this one oral-derived text and had no wider currency in the tradition; nonetheless, the structure of the recurrent scene marks it as traditional. Whether a given unit was a functioning oral traditional element, as in the case of the Serbo-Croatian epic, or the result of the traditional method of composition in a transitional text, patterned recurrency of this sort argues its traditional nature and suggests certain avenues for interpretation.

[30] Cf. Ong 1965; Renoir, esp. 1964, 1976b, 1986, 1988.

There is, of course, no equivalent in the Greek; the deployment of the narrative sequence—a traditional deployment of an oral-formulaic unit in this oral-derived text—supervenes fidelity to source.

In fact, as we move on to the second and third instances of this sequence, we find no equivalent for motif A ("The enemy arrives with a large troop") at any point in the Praxeis . Occurrences #2 and #3 begin with the equivalent of (or source for) motif B, "They lead the prisoner from his cell and drag him about the city":^[33]

Occurrence #2		Occurrence #3
	(Praxeis , chap. 26)		(Praxeis , chap. 28)
Occurrence #2
And in the morning they carried him
out again and, having fastened a rope
around his neck, they dragged him about.

This amounts to further evidence that the Old English scourging theme has a life of its own, even if only within Andreas , and that at times this independent structure supersedes the narrative devil of the source. Here are the Old English passages in question, thee which constitute motif A in the theme and which have no equivalent in the Praxeis :

[32] On the traditional character of bolgenmode ("enraged in spirit"), which names the customary battle-rage that precedes the actual fight, see the analysis of Grendel's approach in chapter 6.

With the exception of a few words like weorud ("troop") that we would expect to find in such a description, these three occurrences of motif A do not reveal much in the way of verbal correspondence. In fact, as but a cursory glance at the Greek versions of motif B shows, even the certainly written and untraditional prose of the Praxeis reveals more word-for-word and phrase-for-phrase correspondence than do these three equivalent passages from Andreas . Thus, while the narrative pattern is without doubt thematic—consistently departing from the source to invoke a traditional motif—this section of the Scourging theme reveals no real verbal consistency .

Before examining certain of the other motifs in this theme in order to determine whether adherence to narrative pattern and lack of verbal consistency prove characteristic of the Scourging scene as a whole, it may be well to provide an overview of the motif-by-motif relationship between the Greek and Old English texts over all three occurrences of the scene. Table 22 makes the correspondence appear more regular than it actually is. In general, the Greek source is much more telegraphic, much less descriptive; furthermore, it couches each motif in about the same language in each occurrence. But its agreement with the Old English text is often no more than nominal, with mere suggestions in the Praxeis giving way to more fully developed motifs in Andreas . In addition, as the line numbers indicate, the Old English poet can

interrupt his narrative pattern to insert details apparently of his own creation, as he does during the first scourging scene with a non-narrative excursus ostensibly meant to mark Andrew as a type of Christ (1250b-52). Nonetheless, the Old English poem typically shows a greater regularity of narrative pattern than does the Praxeis , although verbal correspondence among successive scourgings remains as a rule much higher in the source.

As an example of this set of qualities, we may consider the Old English versions of motif B, whose Greek source was shown above to be almost absolutely consistent. Before examining the passages in question, however, there are two general observations to be made about their relationship to equivalents in the Praxeis . First, the Greek passages, though terse, are not obviously self-contained; while the fact that they open a chapter makes for a clear point of beginning, in the Praxeis motif B flows right into C: without a discernible break in narrative structure. Second, this same motif B ("They lead the prisoner from his cell and drag him about the city"), in taking virtually the same phraseological form each time, does not expand, contract, or divide. In short, it does not participate paratactically in a sequence of actions—as the Old English counterpart most surely does—but seems to be a detail with no particular identity of its own. In contrast, not only is the Andreas motif a self-contained, discrete member of a sequence, but as a structure with some individual identity it can expand, contract, and divide. But although the B motif exhibits this morphology—a morphology typical of Old English themes, unbound as they are by the phraseological constraints that shape thematic structure in ancient Greek and Serbo-Croatian—it nevertheless recurs consistently as a necessary part of the oral-formulaic pattern.

The occurrences of motif B from the Andreas poem are as shown in figure 11. As the parallel passages reveal, aside from the words heron ("they ordered") and lædan ("lead"), once again the narrative element does not take a consistent verbal form. There seems to be room for much variation in the actual diction of the element, just as with motif A. In addition, the poet can tack on an expatiation on the basic action, as he does in lines 1232-38 of the first occurrence; these verses further specify the action of dragging the prisoner through the streets of the city and have no equivalent either in the Praxeis or in the other two occurrences from Andreas . Just as with the Serbo-Croatian theme, apparently, a paratactic addition may be made to the fundamental action of a unit, much in the manner of what Lord terms "ornamentation." We should note that this expansion does not change the traditional structure here (as it did not in the Serbo-Croatian examples in chapter 8), since it simply reinforces and elaborates a traditional idea. In short, what we have in element B is another instance of this Old English theme's consistency of narrative pattern and virtual lack of verbal correspondence.

In lieu of exhaustive analysis of all motifs in the Scourging theme, I shall simply report general results for the remaining motifs (G, D, and E) and men-

tion one telling departure from thematic structure that further illustrates the poet's method. Motif C, "The prisoner's wounds are described," takes (like the unit just examined) a more elaborate form in the first occurrence than in either of the last two. In the second instance it splits into a ring that surrounds Andreas's prayer and the Devil's reply, non-thematic actions from the Greek source that the poet accommodates by enclosing them in a traditional envelope.^[34] This is not an unusual technique in Old English poetry, wherein the pressure toward absolute, one-to-one conformity between thematic idea and phraseological expression is not nearly so great as in Serbo-Groatian or ancient Greek epic. For all of its variations over the three occurrences, however, this section shows a surprising though not overwhelming amount of verbal correspondence. This quality, which by now we must see as uncustomary in Old English, probably results from the greater narrative focus of element C, the actual description of the body wounds suffered by the prisoner; since the action is less generic and more highly specialized, we may conclude that even on a purely semantic (that is, extra-traditional) basis, we could expect more repetition of morphs than in, for example, motif A or B.

Motifs D ("They lead the prisoner back to his cell") and E ("The prisoner confronts night and mental torture") return to the model of the first two substructures, regularly representing essential narrative ideas but having little or no verbal correspondence among instances.^[35] As such, they typify the Old English theme as a whole. But there seems to be one glaring problem; to the essential action of motif E in the first scourging scene (1253-55a)—

[p]a se halga wæs under heolstorscuwan,
eorl ellenheard, ondlangne niht
searopancum beseted.

Thus the holy one was under the shadow of darkness,
the courageous earl, all night long,
surrounded by cunning thoughts.

—is attached a rather long, apparently unrelated excursus (1255b-69a):

Snaw eorðan band              1255
wintergeworpum; weder coledon
heardum hægelscurum, swylce hrim ond forst,
hare hildstapan, hapless eðel
lucon, leoda gesetu. Land wæron freorig;
cealdum cylegicelum dang wæteres prym,              1260
ofer eastreamas is brycgade,
blæce brimrade. Bliðheort wunode
eorl unforcuð, elnes gemynding,

[34] On ring composition, see van Otterlo 1944a, 1944b, 1948; Whitman 1958; Lord 1986a. On the related issue of envelope parterre, see note 15 above.

[35] Note the enclosure of D within E in occurrences #2 and #3.

prist ond prohtheard, in preanedum
wintercealdan niht. No on gewitte blon,              1265
acol for py egesan, pæs pe he ær ongann,
pæt he a domlicost dryhten herede,
weorðade wordum, oððæt wuldres gim
heofontorht onhlad.

Snow bound up the earth              1255
in winter-drifts; the weather grew cold
with hard hail-showers, likewise rime and frost,
gray warriors stalking, locked up men's homeland,
the people's dwellings. Lands were frozen;
water's power hardened into cold icicles,              1260
ice bridged over the ocean-streams,
a shining sea-road. Happy in heart dwelled
the blameless earl, mindful of valor,
bold and patient, in great affliction
through the winter-cold night. Never did he cease
        from reason,              1265
stricken by fear, from what he had formerly begun,
but he ever most gloriously praised his Lord,
honored Him with words, until glory's gem
appeared heaven-bright.

At first sight this second passage seems supernumerary, an unnecessary and perhaps illogical addition to the core action expected in motif E. And while there is a nominal equivalent in the Praxeis for E, it is merely the briefest of phrases (, or "and he was sorely distressed") and has absolutely nothing to do with what follows in 1255b-69. We are left to ask whether this second passage can be viewed as a paratactic expansion of the core action, even though it seems in its lyrical and exacting description of winter's locking up the seaways to go well beyond intensification or ornamentation of motif E. Simply put, what has the quite unexpected onset of winter to do with the prisoner's dark night of the soul? Are we then to see the excursus as a flaw, an unwarranted expatiation rather clumsily tacked on to a motif by virtue of prior association or some other cause? For even if the passage appears quite beautifully evocative in its own right, is it not a departure from the narrative business at hand?

The answer to these questions lies in understanding exactly what the passage represents, and here we penetrate to the innermost layer of the Andreas poet's compositional technique. For the "winter" and "sea" imagery that makes up this cluster of lines is associated in the poetic tradition with the theme of "Exile" first discovered and commented on by Stanley B. Greenfield (1953a,b, and esp. 1955). Although this brief passage, subordinate as it is to another theme, does not develop what Greenfield has identified as the four aspects that collectively typify the Exile pattern in Old English verse, all of the lyric

imagery used in the Andreas instance belongs to (and therefore summons up) the traditional idea of Exile. We need only recall elegies like the The Wanderer, The Seafarer, The Wife's Lament , and more to hear the resonance of the thematic pattern: the isolated, solitary man or woman, deprived of all kinship and social ties, struggles weary-minded to overcome the mental and physical ravages of his or her earthly condition. There is no more powerful or ubiquitous traditional pattern in the Anglo-Saxon poetic corpus, and it comes very near the surface of things as the poet works traditionally to express the tribulations of Andrew's first night in a meaningful way.

That these powerful overtones associated with implicit reference to the theme of Exile are really part of the poetics of this passage can be proved in another way as well. Later on, during the second night of mental anguish, Satan's followers, having first fled Andrew's ready responses, plan to return with their leader to taunt him again. In urging Satan to the task, they say (1358a):

oðwitan him his  wræcsið
Let us mock his [Andrew's]  exile- journey

The use of the morph wræc- , "exile"^[36] —the key word in all instances of the Exile theme—shows how closely the poet identifies the much-persecuted Andreas with the Old English peregrinus figure; the attribution of associated qualities is reinforced when, in the midst of the third scourging, God confronts the holy man with these words (1431-33):

"Ne wep pone  wræcsið , wine leofesta,
his pe to frecne; ic pe friðe healde,
minre mundbyrde mægene besette."

"Do not bewail this  exile- journey, dearest friend,
it is not too fearsome for you; I shall keep you in safety,
in my protection I shall surround you with might."

Not only the use of the key morph wræc- , for which there is again no equivalent in the Praxeis , but also the following lines that, like The Seafarer or The Wanderer , promise or hint at a new, divine context for the lost earthly context, indicate how the Andreas poet conceives of his hero as an exile.^[37] In short, although

[36] From the verb wrecan , "to drive out."

[37] To these two occurrences of wræcsið at 1358a and 1431a we may add other dues that support the identification of the hero with the Exile figure. At 491-92 Andreas speaks of former journeys (that is, trials) using Exile imagery common to the The Wanderer and The Seafarer ; at 889a another instance of wræcsið helps to describe the sinners' journey away from the reward of heaven; at 1283b Andreas describes his tribulations as earfeðsiðas ; at 1351a a demon frustrated by his and his fellows' lack of success in seducing Andreas calls the holy man an anhaga (cf. Wdr 1a, e.g.); at 1380a and 1383a Andreas tells the devil that God could easily cause him everlasting wræc . Even Mattheus refers to himself, the suffering martyr, as an exile (eðelleasum ) at 74b.

the poet may seem to depart from the logical succession of the narrative and to tack on a lyric to a motif already functionally complete (and all this without any prompting from the source), in fact he only turns from one theme to another—from the Scourging pattern to the details associated conventionally with Exile, and back again. In relation both to compositional technique and to aesthetics, this is a consummately traditional shift of narrative structure.

The Relationship Between Phraseology and Narrative Structure

The two examples of the Old English theme examined in this chapter illustrate, each in its own way, the tradition-dependent character of the multiform unit. While thematic structure in Beowulf and Andreas resembles what we have found in ancient Greek and Serbo-Croatian in terms of narrative pattern , the Old English unit lacks the density of verbal correspondence typical of Homer and the epic guslar . There may be a group of morphs associated with certain parts of the Anglo-Saxon theme or occasionally even a classically defined formula, as in the Beowulfian Sea Voyage, but nowhere do we encounter the line- and colon-based correspondence characteristic of instances of themes in the other two traditions. As was remarked above, this lack of regularity in diction is only to be expected in a poetry whose formulaic structure is so divergent from the Homeric and Serbo-Croatian models—that is, whose phraseology operates on such idiosyncratic rules. But, these observations having been made, we must go a step further and inquire how the Old English theme actually does express itself in what has long been known to be a traditional diction. To put the matter directly, given the tradition-dependent nature of the Anglo-Saxon narrative unit, what is the relationship between theme and phraseology?

To begin, we know that this relationship differs from the synergistic but relatively less flexible dynamics of formula and theme in ancient Greek and Yugoslav epic, for in Homer's and the guslar 's craft we discern a number of factors that closely marshal the expression of narrative ideas. As Milman Parry long ago observed of Homer, and as Albert Lord has argued is true (to a slightly lesser degree) for the guslar ,^[38] poets in these two traditions have fashioned a phraseology remarkably free of synonymous, metrically equivalent formulas. These dictions are characterized, in other words, by a high degree of thrift . In practical terms, thrift means that the essential ideas of a traditional poetry approach a one-to-one relationship with their epitomized expressions, so that a poet wishing to express a given essential idea will, in a given metrical format, turn most often to a single expression.^[39] This emphasis on thrift does

[38] See, e.g., Lord 1960, 49-53.

[39] This formulation must of course take into account the poet's use of the formulaic language as an idiom. He does not proceed colon by colon, but "fluently," as it were. Thrift applies most clearly and regularly to noun-epithet formulas.

not of course indenture the poet as slave to the limited resources of his tradition; as Parry began to point out, and as has since been shown at greater length (Foley 1984b, 1986a,b), these expressions are in effect metonymic of the great resources of tradition as a whole. But, the echoic riches of the diction aside, a thrifty phraseology means that a single traditional idea will necessarily lead regularly (even inevitably) to a single expression in the Homeric or Serbo-Croatian Kunstsprache .

Thriftiness stems from the inherent morphological conservatism of ancient Greek and Serbo-Croatian formulaic structure. Were it not that these two phraseologies existed symbiotically with two meters that, as was demonstrated in earlier chapters, encouraged the formation and maintenance of encapsulated, right-justified formulas with regular syllabic definition and internal structure, thrift would be impossible. For thrift depends directly on consistency of diction over time, and such diachronic consistency can obtain only when the institutionalized constraints on diction establish—and prevent the collapse of—well-defined metrical categories. If Homeric formulas, for example, could expand grossly in syllabic count, operate without true caesuras and diareses, and ignore the Indo-European rule of right justification, they would quickly lose their ability to recur regularly and predictably. And this change in metrical constraints would in time lead to a lack of (classical) definition as formulaic units. Furthermore, because uniqueness would perish along with definition, the same process would lead to lack of a simple one-to-one correspondence between essential idea and phraseological unit. In short, what we call thrift would be impossible to maintain.

But in fact that is precisely the case in Old English. While we have no reason to believe that the Anglo-Saxon compositional idiom was ever structured more like its Homeric counterpart,^[40] it is quite clear that nothing like thrift characterizes Old English as it stands.^[41] It would be more accurate to say that both the nature of phraseology and the paratactic, adding style called "variation"^[42] that typify Old English poetic style actively discourage thrift and promote a one-to-many relationship between an essential traditional idea and its verbal expression .

In chapter 6 it was shown that Old English poetic phraseology is much more flexible than either of its comparands: lacking the metrical requirements of syllabic count, regular caesura, and right justification, it manifests instead a formulaic structure that depends on stressed positions and alliteration between half-lines. This arrangement results not in the highly regular (because encapsulated and internally regulated) diction of Homer and the guslar but rather in a far more flexible idiom that permits much more extensive variation around a relatively fixed core of one or more words or morphs. Such a tradition-

[40] On the prehistory of the Old English alliterative line, see chapter 3.

[41] See further Fry 1968c.

[42] For discussions of variation, see, e.g., Brodeur 1969, 39-70; Robinson 1985.

dependent phraseology naturally exhibits a less rigidly governed morphology: formulas as elements do not recur verbatim or with only a strictly limited number of variations; instead, they change significantly from one instance to the next. Indeed, with variability as relatively unregulated as it seems to be, even the concept of "recurrence" itself may not be precise enough to describe the protean adaptability of Old English diction. To sum up, with stress position, sequence of stresses, and alliteration as the primary prosodic features of the Old English line, we cannot expect the kind of recurrence found in Homeric and Serbo-Croatian epic.

If, in turn, that sort of recurrence is impossible under the constraints of Anglo-Saxon prosody, then the tradition could never develop anything approaching a one-to-one correspondence of phrase and essential idea.^[43] And if that correspondence could not evolve, neither could thrift. Under Old English metrical conditions, one idea may have many phraseological correlates rather than a single epitomized expression, for that is the nature of the idiosyncratic prosody that supports phrase-making and -retention. Correspondingly, we have long known that Old English style leans heavily on the poetic device of variation , the terracing of subjects, verbs, or other elements in an additive series. The celebrated short lyric "Cædmon's Hymn" provides an expedient example of this common technique (West Saxon version, ASPR , 6:106):

Nu sculon herigean  heofonrices weard ,
meotodes  meahte and his modgepanc,
weorc wuldorfæder , swa he wundra gehwæs,
ece drihten , or onstealde.
He ærest sceop eorðan bearnum
heofon to hrofe, halig scyppend ;
pa middangeard moncynnes weard ,
ece drihten , æfter teode
firum foldan, frea æmihtig .

Now we must praise the  heavenly kingdom's guardian ,
the measurer's  might and his spirit-thought,
the glory-father's  work, as he of each wonder,
eternal lord , established the beginning.
He first made for the children of earth
heaven as a roof, holy creator ;
then the middle-earth  mankind's guardian ,
eternal lord , afterward appointed
for the people of earth,  lord almighty .

The italicized words and phrases (all of which, except in lines 2 and 3, are exactly one half-line in length) all name God, the creator of the universe, in

[43] The situation should not be phrased only negatively; compositionally, this dynamic also allows for a useful flexibility in verse-making.

his many and diverse aspects, and part of the poem's power derives from the panegyric series of these epithets, one after the other. In the second part of the hymn, four different phrases—"holy creator," "mankind's guardian," "eternal lord," and "lord almighty"—comprise an extended figure of variation. These four epithets, formulaically unrelated, have an additive or aggregate force greater than the sum of their individual contributions because they occur in what may meaningfully be called a rhetorical series, with the last three in direct apposition to one another. Stylistically, the fact that ece drihten ("eternal lord") occurs twice in this poem and many other times in the poetic tradition at large, or that mancynnes weard ("mankind's guardian," 7b) can be construed as sharing a formulaic system with heofonrices weard ("heavenly kingdom's guardian," 1a),^[44] is not so significant as the fact of their co-occurrence in the typically Anglo-Saxon figure of variation.^[45]

Taken together, these observations about (a) the inherent flexibility of the Old English poetic Kunstsprache and (b) the marked stylistic preference for the rhetorical figure of variation lead us to expect exactly what seems to characterize themes in the Old English tradition: they reveal a definite, consistent narrative pattern but little or no verbal correspondence. If the essential ideas embodied in the narrative design have no consistently focused, one-to-one relationship with the elements of traditional diction that serve as their expressive medium, then the theme simply cannot recur with formulaic repetition marking its various instances. Rather, we find the "deep structure" of narrative pattern, to use Calvert Watkins's formulation,^[46] imaged in an unbound, non-specialized set of traditional phrases—that is, in a surface structure of diction that can shift widely in its formulaic and morphemic make-up while staying within the theme's ideational boundaries. The flexibility of Old English poetic phraseology—a multiform surface that is not subject to the law of thrift—makes for not simply a single characteristic rendering of a narrative idea but rather for many equivalent possibilities for expression of the theme. If by "traditional" we mean a phraseology that, no matter how repetitive it may be, provides the poet with a language both sufficiently plastic and sufficiently structured to serve his needs as a compositional tool, then the Old English poetic idiom is fully as traditional as its Homeric and Serbo-Croatian counterparts.

Thus it is that tradition-dependence becomes most crucial as one confronts the relationship between formulaic phraseology and narrative structure. Since

[44] See Fry 1974 for a thorough formulaic analysis of "Cædmon's Hymn." I should add here that I chose the hymn as a convenient example of the figure of variation, not as a comparand for Beowulf and Andreas .

[45] Indeed, it becomes clear that the versificational and stylistic figure of variation actually promotes unperiodic enjambement. As we discovered in chapter 6, the solution to compositional problems often extends over more than one line, supported by the alliteration of half-lines.

[46] As articulated in G. Nagy 1979, 2-6.

we have found in earlier chapters that comparison of formulas or themes across traditions is misleading unless we first take into account the inevitable differences that spring ultimately from discrepancies among natural languages themselves, so we should expect corresponding differences in this important relationship. The following list displays schematically two general models for this interaction.

Homeric and Serbo-Croatian	Old English
Theme (deep structure)	Theme (deep structure)
Bundles of formulas (surface structure)	Unbound selection of formulas (surface structure)

The phraseology and prosody of ancient Greek and Yugoslav epic are such that a one-to-one, bound relationship between the narrative pattern and its formulaic expression can survive over time as—within limits—the single way to capture the traditional wisdom of the theme. Of course, as we have seen, the Homeric and Serbo-Croatian multiforms can also vary in myriad ways on a nominal scale, with addition, deletion, splitting of motifs, and so on; in addition, we have discovered that both of these other epic traditions also include thematic patterns, and parts of patterns, with no firm core of verbal correspondence among their instances. But the phraseology and prosody of Old English are such that the relationship between narrative pattern and phraseological expression is always and everywhere unbound and one-to-many. All phraseology used to express the theme will still be traditional (in terms of the special meaning of that term for Old English verse), but the deep thematic structure will not exert as constraining and deterministic an influence on choice of diction as it can in ancient Greek and Serbo-Croatian epic. It is no accident, in other words, that we do not find occurrences of the "Hero on the Beach," "Exile," or any other Old English multiforms sharing a common fund of formulaic diction, nor is that "lack" a sign that these narrative patterns are not "true themes." The solution to the quandary is once again found in the principle of tradition-dependence, in this case specifically in the idiosyncratic relationship in Old English poetry between the traditional ideas embodied in the narrative pattern of the theme and their actual verbal expression in traditional phraseology.