5. The Formula in Anglo-Saxon Poetry*

In this chapter I explore the degree to which oral traditional style informs, that is to say, manifests itself in, Beowulf and other Anglo-Saxon poetry. The adaptation to Anglo-Saxon poetics of Milman Parry's definition of the formula and the formulaic system, as well as the concept of the "theme" (the latter treated in Chapter 6) have received considerable attention. Much has been written on the formula in Old English, and there have been excellent summaries. One thinks especially of Ann Chalmers Watts's The Lyre and the Harp. [1] We come back to Larry Benson's assertion that written Old English is as formulaic as what might well be oral traditional poetic texts. [2] He was criticizing my application, in The Singer of Tales , of Parry's view of the formula to Beowulf . John D. Niles in a 1981 article and in his 1983 book Beowulf has responded both to Watts and to Benson. He comments on Watts's work as follows:
In particular, the main contention of Ann Chalmers Watts in her study The Lyre and the Harp—that when closely scrutinized, Anglo-Saxon narrative poetry is not highly formulaic and therefore has no claim to be called oral—is unjustified because it is based on a concept of the formula as a fixed phrase … To Watts, a formula in Old English is "a repeated sequence that fills one of {117|118} Sievers' five basic rhythmical types" (p. 90), or in other words, a repeated verse. Watts correctly concludes that the formula thus defined is not of outstanding importance in Beowulf, but since her definition is ill chosen, her conclusion has no bearing on the question of the validity of the oral-formulaic theory. [3]
Niles analyzes the systems of the first fifty verses of Beowulf and then gives the results of a similar analysis of the first five hundred verses. He concludes:
Close to two out of three verses in the poem are members of one or another identifiable formulaic system. In creating these phrases, the poet not only was working within the linguistic patterns afforded him by the natural language, he was using more highly specialized patterns which enabled him to compose fluently in the alliterative form. On the other hand, only about one verse in six recurs elsewhere in substantially the same form. To call these verses "fixed formulas" would be misleading, for most of them (almost 80%) are members of one or another flexible formulaic system. In other words, the diction of Beowulf is indeed highly formulaic, but far more important than the repetition of fixed phrases is the substitution of one verbal element for another within flexible formulaic systems one half-line in length. [4]
Niles carefully restricts himself to Beowulf rather than considering all the poetry in the Anglo-Saxon Poetic Records (ASPR).He criticized certain of Benson's conclusions as follows:
By comparing a number of passages of Anglo-Saxon poetry, Benson shows that formulas occur in the work of authors known to have been composing pen in hand. He concludes that since all Old English poetry is formulaic, there can be no way of determining the oral vs. written nature of a given text. Benson has well shown that blind formulaic analysis of a particular passage of Old English poetry is no sure proof of its mode of creation. Many a learned author has been known to use formulas, whether for rhetorical effect, for atmosphere, or because the poetic language itself is steeped in them. In the absence of definite information indicating the mode of creation of a text, one must use great caution in attributing it to a singer. All the same, like Watts, Benson fails to make a distinction between living formulaic language and the parroting of formulaic tags. In seeking evidence for the mode of composition of a text, one must not only count the number of fixed formulas in the work, for these are easily imitated by a lettered author. One must look for evidence that the poet in question made formulaic language his habitual mode of thought . [5] {118|119}
My reply to Benson's article was "The Formulaic Structure of Introductions to Direct Discourse in Beowulf and Elene," [6] in which I was able to show that the formulas used in introductions to speech are handled by the Beowulf poet in a way different from that of any other Anglo-Saxon poet whose works we have. The maþelode '(he) spoke' formula systems belong to Beowulf par excellence; in fact "they are used twenty-six times in Beowulf compared with nine in Elene and four in Genesi s—twice as many in Beowulf as in the other two combined." [7] But that is not all, as I pointed out:
It is not, however, merely that the formula systems on maþelode are used more frequently than in any other system in Beowulf or than elsewhere in Anglo-Saxon poetry. What is more significant is that they are used differently. The poet of Beowulf does not hesitate to employ these formulas over and over again, even in long sequences, without striving for variety, although we can see from the list just given [of other words to introduce speech] that he had alternatives in his repertory, if he chose to use them … It is especially significant to observe the two "runs" of eight or nine maþelodes [between lines 348 and 631 and between lines 1215 and 1999]. They, as well as the word itself, differentiate Beowulf from the other Anglo-Saxon poems. Within the poem they also appear to distinguish the scenes in Heorot from that in Hygelac's hall and to show a continuity with the dragon episode, marking significantly the generational succession from Beowulf to his kinsman Wiglaf. [8]
In Cynewulf's Elene, however, in long series of introductions to direct discourse such as between lines 78 and 537 and between lines 604 and 685, with a number of maþelodes, the use of this verb is interrupted by other words for introducing speech. A noteworthy example of how the series of words for speech is broken occurs in the second major group of exchanges of speeches in Elene, beginning in line 573 with Elene's address to the assembled people. As I have written,
this fact is the more amazing because in the passage in question Cynewulf is following the Latin source very closely, and the Latin speeches are all, without exception, introduced by dixit. Cynewulf varies the verbs of speaking more than either the Beowulf poet or the Latin does. In addition Cynewulf in this passage employs the epithet for Elene used regularly in the Latin text, beata, which becomes eadige in Anglo-Saxon, only once (619) in Elene, Cynewulf also {119|120} uses tireadig (605) and œđele (662). In short, here too as in the sequences of words introducing speech, Cynewulf prefers variety. [9]
It is not enough to say, as Benson does, that Old English poetry is "both formulaic and lettered." [10] It is essential to realize that one can distinguish in number and usage the formulas in Beowulf from those in the Old English poems that stem from religious and from learned secular sources. A formulaic "residue" carried over from a previous oral practice is different from the formulaic systems in a poem that belongs to a still-living heroic tradition.
In his book on Beowulf , Niles adds a consideration of "compound diction" as a mark of the poet's oral style:
One striking feature of Beowulf is its extraordinary wealth of compound diction. To a large extent this vocabulary appears to be formal and inherited rather than idiosyncratic, and it seems a particular adjunct to the heroic style … These compounds have long attracted attention. They have been counted and catalogued, and their aesthetic effect has been described. All I need to stress is that they are not merely ornamental; they are not used merely for poetic effect. They are functional, and this is their prime reason for being. The unusually rich and varied compound diction in Beowulf is the product of a centuries-old tradition of alliterative verse-making, and the poet used it not only because he loved it but because it helped him compose. [11]
In his study of Beowulf, Niles is indebted to Donald K. Fry's definition of the formula, which is the one most commonly accepted by Old English scholars. Fry's most noteworthy and important articles are "Old English Formulas and Systems," and "Some Aesthetic Implications of a New Definition of the Formula." [12] He defines an Old English formula as a "group of words, one-half line in length, which shows evidence of being the direct product of a formulaic system." In Milman Parry's view, groups of related formulas made up "systems" from which Homer chose his diction, but Fry proposed, rather, that "a formula is generated from a system," which he defines as "a group of half-lines, usually loosely related metrically and semantically, which are related in form by the identical relative placement of two elements, one a variable word or element of a compound usually supplying the alliteration, and the other a constant word or element of a compound, with approximately the same distribution of non-stressed elements." [13] {120|121}
Fry summed up his conclusions on formulas and on oral versus written texts in Old English in a 1977 paper published in 1981, and I can do no better than quote parts of its last three paragraphs:
The basic building block of this poetry and of any oral poetry is the formula. But formulas work differently in different traditions. Parry's classic definition of a formula as "a group of words which is regularly employed under the same metrical conditions to express a given essential idea" fits Homeric Greek rather well, but Old English poorly. An Old English formula may consist of one long compound word rather than "a group of words." The formula may appear only once, not "regularly employed," but be closely related to other formulas from the same system. Since all Old English half-lines are metrically equivalent, "under the same metrical conditions" proves meaningless. And finally Old English formulas seem to be related only in form, not in content, i.e., they do not "express a given essential idea." The first step in determining the formula in any given corpus requires identification of the basic unit of composition. In Greek, dactylic feet in various combinations comprise the units. For Old English, the unit is the half-line. The Middle English unit seems to be stressed pairs of words occurring together. [14]
I do not wish to detract from Fry's important contributions in this field; for they moved the study of Anglo-Saxon formulas forward on their own path, and they also broadened and deepened our sense of the intricacy of structure characteristic of oral traditional narrative style. I find it necessary, however, to comment on several aspects of his definitions of formula and system vis-à-vis Parry's.
One, Fry's 1967 definition of the formula, as quoted above, began with "a group of words," but in 1977 he stated that Parry's definition was inadequate for Old English, because "an Old English formula may consist of one long compound word rather than 'a group of words.'" Parry's first definition of the formula was "in the diction of bardic poetry, the formula can be defined as an expression regularly used, under the same metrical conditions, to express an essential idea." [15] The form of the definition I generally use begins "a word, or group of words." The debate as to whether a one-word formula is legitimate or not is of long standing and will probably continue.
Two, Fry's definition of the formula in Old English seems to eliminate two elements in Parry's definition, namely, "regular use," and "essential idea," which I believe are vital to an understanding of the nature of a formula. Fry's definition is mercifully simple, but after placing the formula, {121|122} quite properly, in a system, he then presents us with a definition of a system, which is more complex. Actually, Parry stated that it is not necessary to find a phrase more than once in any text for it to be a formula. He clearly stated, as we saw above, that he was approaching the formula from the side of repetition because it was easier to see that it was used regularly if it was repeated but that its repetition in any given text depended on whether the idea embodied in it was needed again under the given metrical conditions— including "placement in the line," to use Fry's phrase—and not on the nature of the formula itself. [16] There is a crucial difference between a formula in an oral poem and a repetition used by a writer in a literary milieu. Formulas are vital for the oral poet; for they help in verse making. A poet in a written tradition employs repetition for aesthetic effect or for referential reasons. Formulas embody all previous occurrences and not any particular one; in an oral poem they do not point to other uses of the same formula.
"Regular use," however, is important in the definition of the formula, be it in Old English or in any other tradition, no matter how one chooses to word it. If one wishes to interpret Fry's definition of a system as implying "regular use," and as including, under "semantic," the concept of "essential idea," then his definition may not be so radical as it may seem but may be the same thing in disguise, except for the restricting of Parry's "metrical conditions" to a half-line in the case of Old English. Be that as it may, Fry's discussion of his definition in his 1967 article admirably analyzes what I have later termed the "weaving style," namely, the intricate pattern of interrelationships of formulas and of formulaic expressions which is typical of oral traditional style.
Three, in limiting the formula to a half-line in the case of Anglo-Saxon poetry, Fry has, I believe, been too conservative. Recent work seems to indicate that we should extend the boundaries of the formula because, like everyone else, the singers do not think, or compose, in terms of half-lines. They think in whole lines or, even better, in terms of clauses, or perhaps of whole sentences, which may well be of greater length than a whole line, to say nothing of a half-line. In other words, they do not compose in terms of subjects or of verbs but of subjects-plus-verbs, that is, in terms of larger syntactic and semantic units. If we leave untouched Fry's restriction on length for the Old English formula, we will have to invent another simple term for the longer units.
Fry also expressed his opinion on oral versus written Old English poetry: {122|123}
For Old English the hardest problem turned on identifying oral versus written texts … Recently most scholars believe that oral and written texts are indistinguishable anyway, since Old English writers inherited and imitated the formulaic style, indeed the only poetic they knew. We can distinguish oral texts only when some contemporary, such as Bede, labels them for us; unfortunately, for Old English, that includes only Cædmon's Hymn. Certain factors in medieval authorship complicate this picture further. Anglo-Saxon poetry lacks a chronology. Fewer than ten Old English poems can be dated at all, and we think Beowulf was composed sometime between 658 and 1025! Only six poems have known authors; indeed the whole idea of authorship or literary property was different or perhaps non-existent in the vernacular. Poets simply recomposed each other's works without attribution or guilt. [17]
Fry thus states that the method of composition is the same whether it be done orally or in writing. There is merit in that suggestion, I believe, but only for the earliest period of written Old English poetry, the first stage of transition, as it were, or the last of oral traditional poetry, depending on the direction of one's approach. It should be our task to determine how the oral formulaic style changes in time as the method of composition changes; for change it must, and does, through the increased influence of reading and writing and, more particularly, through the eventual imitation of new poetic models, for example, Latin poetry. Those changes in style are visible and measurable.
Fry's final paragraph deals with aesthetics:
Finally and paradoxically, although we cannot distinguish oral from written texts, oral aesthetics are different, understandable only on their own terms. For example, repetition in a writing poet is a flaw, but for an oral poet repetition is a badge of loyalty to inherited poetic traditions and values. Oral poems "sound alike" internally and externally because their poets wanted them to, to please an audience appreciative of traditional verse. [18]
Fry is aware of the inconsistency, as he notes the paradox, but he makes no attempt to explain it. Yet it cries out for explanation. Would not the loss of repetitions of which he speaks as distinguishing marks of oral aesthetics be visible and measurable when the aesthetics change, when the repetitions begin to be less and less acceptable, thus modifying the method of composition? {123|124} When does that begin to be perceptible? There are many questions still to be investigated.
Anita Riedinger's article "The Old English Formula in Context" builds on the work of Fry. [19] She redefines for Old English the terms formula and system, which have been current in scholarship since the days of Milman Parry, and gives a definition for a "set," a word used by Valerie Krishna and by John Niles for a "system." [20]
Riedinger's idea of content is explained simply by observing the situations in which a formula such as nihtlangne fyrst 'the space of a whole night' is used. It occurs five times in four different poems: once each in Beowulf, Exodus, and Elene, and twice in Andreas. In all five cases it "signifies a terrifying period of time prior to a battle." To substantiate this statement she notes that "Unferth dares Beowulf to wait that long near Grendel (Bwf 528a)"; "the terrified Israelites … wait nihtlangne fyrst for battle with the Egyptians (Exo 208b)"; "the fearful Constantine and his Romans ... do the same before battle with the Goths and Huns (Ele 67b)"; the angels who transported Andreas to the land of Mermedonians leave him asleep "nihtlangne fyrst outside the gates of his enemies (And 834b)"; and he waits in prison that period of time before being returned for torture. [21] "In each instance," she says, "the context reveals ominous connotations not explicitly suggested by the formula." To her, this formula always carries with it such "ominous connotations." Such formulas "may be said to signify themes rather than to express them, because they are dependent for their full meaning upon a context external to the semantics of the formulas themselves; that is, the words themselves do not mean 'impending disaster,' yet this is their invariable context." [22] I agree with her that, if this is true, such a meaning would be traditional, because it is shared by more than one poet. I would add that it is likely that considerable time would be required for such a meaning to become traditional. It follows, then, that traditional "thematic formulas" must be of some age. I suggest that they, like the basic formulaic style itself, {124|125} must also have originated in the period when the poetry was oral rather than written.
Riedinger expands the context of the thematic formula by noting that some formulas of that type "stand … at the center of a cluster of recurrent motifs that are not themselves usually expressed by the same formulas." [23] Thus wan under wolcnum 'dark beneath the clouds' is used four times, once each in Beowulf, The Dream of the Rood, Guthlac B, and Andreas. "The cluster of recurrent motifs" in this case are "death or sleep, a shining light, and shadows." In The Dream of the Rood the words of the Cross, speaking of the crucifixion, make use of these motifs. (The underlinings are mine.)
                                     þystro hæfdon
bewrigen mid wolcnum ǁ wealdendes hræw,
scirne sciman, ǁ sceadu forðeode,
wann under wolcnum. ǁ Weop eal gesceaft,
cwiðdon cyninges fyll. ǁ Crist wæs on rode.
DrR 52b-56
Darkness had covered the Ruler's corpse, shining radiance, with clouds; a shadow went forth, dark beneath the clouds. All creation wept, mourned the King's death. Christ was on the cross.
We find the formula in Guthlac B in the description of the night when the saint is dying:
                                     Þa se æþela glæm
setlgong sohte. ǁ swearc norðrodor
won under wolcnum, ǁ woruld miste oferteah
þystrum biþeahte, ǁ þrong niht ofer tiht
londes frætwa. ǁ Ða cwom leohta mæst,
halig of heofonum ǁ hædre scinan,
beorhte ofer burgsalu. ǁ Bad se þe sceolde
eadig on elne ǁ endedogor,
awrecen wælstrælum.
Glc I278b-86a
Then the glorious radiance reached its setting; the northern sky dimmed, dark beneath the clouds, covered the world with mist, wrapped it in darkness; night rushed across the expanse of the land's adornments. Then, holy from the heavens, {125|126} the greatest of lights came shining clearly, bright above the city's dwellings. He who had to do so awaited his last day, blessed with courage, pierced by death's arrows.
Riedinger continues: "The next ten verses describe the miraculous light that illuminates the darkness and include the phrase scadu sweþredon 'shadows disappeared'" (300).
The example from Beowulf occurs when Hrothgar, aware that Grendel will visit Heorot, departs from the hall to go to bed:
                                     oþþæt semninga
sunu Healfdenes ǁ secean wolde
æfenræste; ǁ wiste þæm ahlæcan
to þaem heahsele ǁ hilde geþinged,
siđđan hie sunnan leoht ǁ geseon ne meahton,
oþðe nipende ǁ niht ofer ealle,
scaduhelma gesceapu ǁ scriðan cwoman,
wan under wolcnum.
Bwf 644b-51a
… until presently the son of Healfdene wished to seek his evening rest. He knew the battle planned by the terrible one for that high-hall after they could not see the sun's light and night, obscuring all, the shapes of covering shadows, came gliding, dark beneath the clouds.
Neither the word sleep nor the word death is used specifically in the Beowulf passage; sleep, at least, is implied in œfenrœste 'evening rest'. Moreover, a little later in the poem, when Grendel approaches Heorot, the poet uses some of the words belonging to the thematic cluster:
                                     Com on wanre niht
scriðan sceadugenga. ǁ Sceotend swæfon
Bwf 702b-3
In the dark night the walker-in-shadows came gliding. The warriors slept
Twelve lines later, a variant of wan under wolcnum is found with the shining brightness of gold, a "light" not noted by Reidinger, but which I deem may be pertinent to the theme in this part of Beowulf:
Wod under wolcnum ǁ to þæs þe he winreced, {126|127}
goldsele gumena, ǁ gearwost wisse,
fættum fahne.
Bwf 714-16a
He strode beneath the clouds, until he might most clearly perceive the gold-hall of men, the wine-hall shining with gold plate.
I see no reason why Riedinger should feel that wod under wolcnum is "non-traditional," purely on the grounds that it is not found again in the corpus. I prefer to view it simply as a formulaic expression, possibly a formula, that is an apt variant of wan under wolcnum.
Finally, wan under wolcnum is found in Andreas in the same context where we previously noted the thematic formula nihtlangne fyrst, namely, after the angels have carried Andreas miraculously during a sleep to Mermedonia and left him still asleep for the remainder of the night outside the walls of the city where his enemies dwell. Here is the passage, including the coming of dawn:
Leton þone halgan ǁ be herestræte
swefan on sybbe ǁ under swegles hleo,
bliðne bidan ǁ burhwealle neh,
his niðhetum, ǁ nihtlangne fyrst,
oðþæt dryhten forlet ǁ dægcandelle
scire scinan. ǁ Sceadu sweðerodon,
wonn under wolcnum.
And 831-37a
They let the holy one sleep in peace by the army road under the protection of the sky happily to await near the city wall of his enemies, for the space of a night, until the Lord allowed the day-candle to shine brightly; the shadows disappeared, dark beneath the clouds.
Riedinger interprets this passage according to the other contexts in which nihtlangne fyrst, on the one hand, and wan under wolcnum on the other, appear in the corpus. I am sorry to say that I do not entirely agree with her, tempting as it may be to see these thematic formulas casting an ominous gloom on this passage. There just are not enough cases in Anglo-Saxon poetry to prove that the context is always ominous. She has, indeed, indicated that it is not. She has demonstrated, however, that a cluster of formulas joins these four or five scenes in these poems, and she has noted that in all but one of them, the context is gloomy and ominous. That is interesting and valuable. When she stretches the evidence to create an absolute, she does disservice to her cause. One must be careful about such generalizations; for the body of extant Anglo-Saxon material is limited and varied in genre. {127|128}
The movement to expand the idea of the formula is not actually new. Parry's systems of formulas, as he elaborated them for the Homeric poems in 1928, could be thought of as expansions of the formula. The "formulaic expressions" that I discussed in general and described in detail for both South Slavic and Old English in The Singer of Tales, being based on the idea of Parry's "systems" of formulas, could also be considered expansions of the concept of the formula. [24] It is true, however, that we distinguished the formula from both the system, a group of formulas sharing a syntactic pattern and at least one word, and also from the formulaic expression, which might be thought of as a member of a system, or a partial repetition, a phrase sharing at least one word with the other phrases. It is also true that we considered the formula itself to be a verbatim repetition. In a seminar on medieval epic and romance that I used to offer regularly in the Department of Comparative Literature at Harvard, we accepted as formulas repetitions in which two words were metathesized, and we also counted as formulas phrases in which the elements of formula were declined or conjugated, taking into account necessary adjustments in metrical conditions. Riedinger calls such cases, I think often correctly, the "same formula." For example, she considers hwœr ic under swegle (Wids 101a), "where beneath the heavens I," and nœnigne ic under swegle (Bwf 1197a), "Of none I beneath the heavens" to be the same formula, although I would prefer to think of them as aspects of the same formula. [25] The poet of Andreas uses ofer cold wœter, ond on cald wœter, and on cald wœter, and I find no difficulty in calling all three expressions the same formula, although when Riedinger adds ofer cald cleofu, "across the fateful cliffs" (And 310a), I prefer to note that the phrase is a member of a system with ofer cald as its stable core and wœter and cleofu as the variables, before agreeing to call them the same formula. I hesitate, because these distinctions in "fine-tuning" may turn out under certain circumstances to be useful, and one is running the danger of calling everything a formula indiscriminately or on purely subjective or intuitive grounds.
Riedinger does make an important distinction between an "incidental" and a "significant" repetition. The first is determined by prosody and syntax, the formula, in short, with which we are familiar; the second is "intentional—the deliberate repetition of a verse such as ofer cald wœter or wan under wolcnum in order to convey a specific meaning." [26] The significant formulas are thematic formulas. They imply, or signify, a theme; receive their significance {128|129} from their traditional contexts; and are used by the poet in order to evoke the essence of those contexts.
There are at least two other kinds of expansion—or directions in which expansion may go—that should be mentioned here. One is by making the formula longer, that is, by extending it beyond the half-line and the whole line to encompass two and more lines. Another is contained in what I have called the "weaving style" of oral traditional poetry. Tod Luethans has described the interlocking of formulas both in the Old French chanson de geste Gormont et Isembart and in the Chanson de Guillaume. [27] A noun-epithet formula can have systems involving the noun and other systems involving the epithet. Each noun listed in the system based on the epithet may have its own system of epithets; each epithet listed in the nominal system may in turn have a system of its own, and so forth. Formulas are thus interrelated through such overlapping, or interwoven, systems.
Those are but two ways, however, of "expanding" the formula, and scholars in Old English have been moving forward very provocatively in others as well. Similar expansions also have taken place in other language areas, particularly Homeric studies, beginning with the work of Notopoulos, Hainsworth, and Hoekstra, and continuing with that of Nagy and Janko. [28]
Benson did not try to redefine the formula but accepted Parry's definition for application in Old English. Before Riedinger, the greatest modification of the definition of the formula had come from Fry. It seems that all scholars who have concerned themselves with the subject have thought of Parry's definition as not applicable to Old English, although it was applicable to Ancient Greek and South Slavic. I believe that it is time to point out that Parry's definition is not specific to either Greek or South Slavic. Consider each phrase in the definition: "A group of words" is appropriate to any language that uses words, and certainly Old English does. "Regularly used" is certainly general enough—it implies that the group of words is generally used in the practice of the singers in any given tradition; it does not say it must be used such-and-such a number of times. "To express a given essential idea" is also not language specific—the essential idea does not have to be Greek, or Slavic; as long as the Anglo-Saxons expressed ideas they were not excluded.
It is "under given metrical conditions" that has caused the difficulty. For {129|130} some unknown reason it has been assumed that the phrase meant either dactylic hexameter or South Slavic decasyllabics. Inasmuch as Old English poetry is in meter, Old English scholars should be able to translate the phrase into the metrical system of Old English, even as Greek scholars translate that general, nonspecific metrical term into dactylic hexameters, or South Slavs into asymmetric decasyllabics. Russian scholars have found no problem in accommodating it to the tonic verse of the byliny, and Old French scholars have not found the assonance of the laisse difficult to include in the umbrella term of "metrical conditions." In short, surely Old English poetry has "metrical conditions," or Eduard Sievers, M. W. M. Pope, Robert Creed, and John Foley have lived in vain. I see no reason to change Parry's definition for the formula. I have already treated Fry's valiant attempts to define the Old English formula as the product of a system while limiting the formula to a half-line. Saying that a formula is the product of a system is not to define it but merely to say whence it came; and to limit it to a half-line is simply to read into Parry's generic definition the specific "metrical conditions," or part of them, that Fry sees as appropriate for Old English. Parry's definition still stands for any poetry that uses words and has some form of "metrical conditions." Each language area will set forth its own particular "metrical conditions," but there is no need to change the definition.
With that much behind us, let us look at Riedinger's concept of formula and system, because, although we may have problems of terminology, I believe that her distinctions are worthy of consideration. She defines a formula thus: "The repetition of one general concept + one system + one function = one formula." [29] Parry's "essential idea" has been replaced by "general concept." Her example is x under (heavens). "Under" is constant; "heavens" may be expressed by several words, such as swegle, wolcnum, roderum, and so forth; and x may be anything, noun, pronoun, adjective, or adverb. The system involved in the example is x under x. If I understand Riedinger correctly, there are two functions possible. One is as a "tag," a way of making a line; the other function is a thematic formula, as described above. Rume under roderum, "spacious beneath the heavens," is a tag; in Phoenix 14a "the verse describes the forests of paradise," and in Genesis 12433a it "describes how the descendants of Noah multiplied." "The two share no contextual meanings, and therefore do not function as thematic formulas," she points out. The repetition of rume under roderum is "incidental." It is a tag the function of which is "chiefly prosodic." But wan under wolcnum, as we have seen in detail above, functions as a thematic formula. {130|131}
Riedinger's definition of a system is admittedly derived from Fry's and runs as follows: "A system is a group of verses usually sharing the same meter and syntax in which one word, usually stressed, is constant and the other stressed word or words may be varied to suit the alliterative and/or narrative context" (305). In Anglo-Saxon terms, this definition of a system is not much different from Parry's, or from mine, for that matter, and it is close to Fry's. There is nothing remarkable in it. But from it she derives another grouping of formulas, which she calls a "set" and defines as follows:
A "set" may be defined as a group of verses usually sharing the same function and system in which one word, usually stressed, is constant, and at least one stressed word may be varied, usually synonymously, to suit the alliterative and/or narrative context. A system may contain several different sets, each of which is a different formula, but all the verses in a set constitute the same formula—whether or not they repeat one another verbatim. By making a distinction between verses that are members of the same system, but of different sets, one can identify word groups that are the equivalent of the "same formula," even though they contain variables. (306)
This is all somewhat confusing and perhaps unduly complex, but it is to be noted that a formula and a set share the elements of function and system. A formula is a repeated general concept + system + function. In other words, both a formula and a set belong to a system; and both a formula and a set have a function. For a formula, we saw that Riedinger acknowledged two clear functions, one as a tag, the other as a thematic formula in context. But we learn later that other functions, of a less precise nature, are adduced. Function does not enter the definition of system, and it is in respect to function that set is of significance. It is the component of function of Riedinger's groupings that saves her work from being just another reshuffling of terms for already familiar phenomena. The system and the set according to Riedinger will be clearer if we look at some of her examples.
She illustrates both system and set with an analysis of the system x hremig 'exulting in x'. [30] She finds two of the sixteen members of this system not to be formulas: wuldrum hremge 'exulting in glories' (Chr 54a) and feðerum hremige 'exulting in their wings' (And 864b). Riedinger comments: "They share neither concept nor function with any other members of the system and are therefore not demonstrably formulas." I believe that Fry would call them formulas because they are members of a system, and I would call them {131|132} "formulaic expressions," although I am ready to admit that most "formulaic expressions" are really formulas. (I like to keep some distinction between verbatim repetitions and those with at least one constant and one or more variables.) Both Fry and I, it seems to me, would find her decisions against these two verses too harsh.
The remaining fourteen members of the x hremig system fall into three sets. The first has only one verse, but it is used six times:
Set 1. blissum hremig 'exulting in joys' (And 1699a; Ele 1137a; ECL 64a; Glc 1106b; Phx 126b, 592b).
This verse, Riedinger tells us, "recurs in varying contexts and functions as a traditional expression of Christian happiness. The frequency of this verbatim repetition makes it an easily recognizable formula" (310). It is the function, nonthematic in this case, that makes this formula into a set. It is not clear to me that anything of moment is gained thereby.
The second set is the opposite of the first in respect to meaning. Its three members may be translated "clamorous with grief."
Set 2. geohðum hremig (SB1 9b)
          gehþum hremig (SB2 9b)
          sorgum hremig (SFt 208b).
Riedinger comments that these formulas "function as a traditional expression of Christian lamentation for man's sins" (310). The first two would be formulas according to both Fry and myself, and the third also according to Fry. I am perfectly willing to go along with this judgment, because all three are synonyms. It is worth pointing out that in spite of their having the same meaning, they do not violate the principle of thrift, because they have different alliteration.
The third and last set has five members, and there is somewhat greater variety in meaning:
Set 3. huðe hremig 'exulting in his booty' (Bwf 124a; Ele 149a)
          since hremig 'exulting in his treasure' (Bwf 1882a)
          frœtwum hremig 'exulting in his treasures' (Bwf 2054a)
          wiges hremige 'exulting in battle' (Brb 59b).
According to Riedinger, "Set 3 is a thematic formula whose function it is to signify the theme 'the victor's reward.' It usually appears with formulas expressing the general concept 'to seek home,' so that the complete theme {132|133} may be identified as 'the victor returns home with his reward'" (310). In the first case, Grendel returns to his lair with thirty men from Heorot; in the second, Constantine returns home from his victory over the Goths and Huns; and in the third, Beowulf returns home from Denmark with his rewards. Riedinger has to stretch a point for the inclusion of the last two examples, because their contexts, especially in the third Beowulf example, are not so clearly related to the first three as one might hope. I find no problem in considering the Brunanburh passage with its depiction of the English forces returning victorious, even though they are "exulting in battle" instead of "booty" or "treasure," as belonging to the same theme as the first three. The fourth instance above occurs in one of the digressions in Beowulf and concerns the descendant of the victor exulting in the sword that had been previously won as booty. There is no returning home, except perhaps by implication, but the context is not so far removed from the sense of the other passages to quibble about its traditionality.
Riedinger closes her article with the following comment:
I believe that with further study my definition of a set, which is the same as my definition of a formula—the repetition of one general concept + one system + one function—can be further narrowed and refined; but the key to that greater precision lies in the study of traditional contexts, which allows us to discern what the Old English poets themselves seem to have regarded as formulas. (317)
The principle of contextual analysis is a very worthy one, it seems to me, and should be pursued, as Riedinger urges, with greater precision. There are, of course, pitfalls, of which she is thoroughly aware. We have a limited body of Anglo-Saxon poetry and it is varied in its genres. To jump to the conclusion that a formula used five times in four poems in a given context is used "invariably" in that type of context is assuming a great deal, even though the statement is literally true. Moreover, one has to strain unduly to interpret some contexts as being of the nature that one would like them to be for the theory to fit. Subjective judgment comes into play more than Riedinger admits, albeit she tries hard to be careful and honest. Nor is context the only element in her analyses and definitions which lies open to the charge of "subjective." Her terms general concept and function are not models of precision; yet they are the cornerstones of her definitions of formula and set.
To turn now to another aspect of the formula, we note that the oral formulaic style is characterized by a high degree of "thriftiness," that is to say, there is usually only one way of expressing an essential idea under any given metrical and—for Anglo-Saxon, or Germanic in general—alliterative conditions. {133|134}
Thrift is a rich subject for Anglo-Saxon poetry with its requirements of alliteration and its propensity for apposition. Nevertheless, few scholars have investigated it. One of the exceptions is William Whallon, who began writing about Beowulf in 1961. [31] In his 1969 paper, Whallon suggests that there are some cases of lack of thrift in Beowulf. He cites two noun-epithet phrases for the hero which have the same alliterative value, rof oretta 'the brave warrior' in line 2538 and rices hyrde 'the guardian of the kingdom' in line 3080, and two similar phrases for Providence, ece drihten 'the everlasting Lord' in line 108 and ylda waldend 'the Ruler of men' in line 1661.
Superficially it is true that rof oretta and rices hyrde have alliteration in r and each contains four syllables. But the distribution of long and short vowels and of stresses is different. Rof oretta has two long os, with stresses on both and a possible weaker stress on ett, whereas rices hyrde has a long i, which is stressed, and a short y, which is also stressed. Thus the two verses are not equivalent except in number of syllables and in alliteration. The differences in distribution of long vowels and of stresses is important and should not be ignored. Moreover, rof oretta is found in the b-verse, whereas rices hyrde is used twice in Beowulf, and only in the a-verse, and belongs in a larger system that includes folces hyrde—used four times in Beowulf in the a-verse (610, 1832, 1849, 2981), hordes hyrde—once in the a-verse (887), and others. Only once does the Beowulf poet use a member of this a-verse system in the b-verse; that is folces hyrde in line 2644. Oretta is used only twice in Beowulf, once in the a-verse (yrre oretta, line 1532) and rof oretta in line 2538, where it alliterates with ronde in the a-verse, Aras ða bi ronde 'arose then by his shield'. To follow that a-verse with rices hyrde would break the poet's usage of that essentially a-verse formula. One should also note that ronde and rof share more than an r; they also share the following o. The poet is thrifty here.
As for ece drihten in line 108, and ylda waldend in line 1661, the story is much the same, although it is true that the distribution of stresses as well as the alliteration (or assonance in this case) is the same in both phrases. All the vowels in the two formulas are short, however, except for the first e of ece,which is long. It would be well to see the context of the two lines in question before we move further.
siþðan him scyppend ǁ forscrifen hæfde
in Caines cynne. ǁ þone cwealm gewræc
ece drihten, ǁ þæs þe he Abel slog;
. . . . . . . . . . . . . . . . . . . {134|135}
Ne meahte ic æt hilde ǁ mid Hruntinge
wiht gewyrcan, ǁ þeah þæt wæpen duge;
ac me geuðe ǁ ylda waldend
þæt ic on wage geseah ǁ wlitig hangian
eald sweord eacen ǁ (oftost wisode
winigea leasum), ǁ þæt ic ðy wæpne gebræd.
Lines 106-8 and 1659-64
… after the Creator had condemned them. On Cain's kindred did the everlasting Lord avenge the murder, for that he had slain Abel … I could do nothing in the fray with Hrunting, trusty though that weapon be. Howbeit the Ruler of men granted me that I might see hanging in beauty on the wall a huge old sword (often and often has He guided those who are deprived of friends), so that I drew that weapon. [32]
Once again the two phrases inhabit different verses, one the a, the other the b. Ylda waldend would have been impossible in the a-verse of line 108, and ece drihten, while just possible perhaps, would be really unhappy in the sea of ws, which start at line 1660 and roll on through line 1664. Actually, the Beowulf poet uses wuldres waldend three times in the a-verse (lines 17 [wealdend], 183, and 1752). Whallon has chosen too hastily. If anything, he has strongly confirmed the thriftiness of the Beowulf poet.
Let us take another instance, namely, the line Beowulf maþelode, / bearn Ecgþeowes, 'Beowulf spoke, the son of Ecgtheow' with which we are already familiar, and examine it from the point of view of thrift. [33] Are there other ways in which the Beowulf poet says "Beowulf spoke" in one line with b-verse alliteration? Beowulf maþelode ǁ bearn Ecgþeowes is found nine times in the poem. Beowulf maþelode with some other b-verse occurs three times, the first time in lines 405-6:
Beowulf maþelode ǁ (on him byrne scan,
searonet seowed ǁ smiþes orþancum):
Beowulf spoke, the corslet on him shone, the armour cunningly linked by the skill of the smith.
When Beowulf first appears before Hrothgar in Heorot, the poet presents him to Hrothgar, and to us, in his shining byrnie in line 405b, followed by a typical appositional line, which is also, by the way, a good example of the adding style of oral traditional verse making, of unperiodic enjambment, to {135|136} use Parry's terminology. In this case "Beowulf spoke" is limited to the a-verse, and a new idea begins in the second half of the line.
There are two instances, however, in which the poet uses an appositional clause in the b-verse referring to the verb rather than to Beowulf. The first is in the introduction to Beowulf's final boast before fighting with the dragon (lines 2510-12):
Beowulf maþelode, ǁ beotwordum spræc
niehstan siðe: ǁ "Ic geneðde fela
guða on geogoðe; ǁ gyt ic wylle …"
Beowulf discoursed—spoke a last time with words of boasting:—"I ventured on many battles in my younger days; once more will I, …"
Here again is a classic example of unperiodic enjambment. Were we to stop at the end of line 2510, we might have thought that here was an instance in which the poet was not being thrifty. But in the b-verse he introduces an idea that continues into the a-verse of the following line. One simply cannot stop at the end of line 2510, because the significant and emotion-filled words, niehstan siðe 'for the last time' come in the following line. Besides, beotwordum spræc introduces a new idea, that of boasting, in the b-verse. The appositive is not a colorless repetition of the verb of speaking but a meaningful addition.
The second instance of an appositive to the verb maþelode in the a-verse is in lines 2724-25:
Beowulf maþelode ǁ (he ofer benne spræc,
wunde wælbleate; ǁ wisse he gearwe,
Beowulf discoursed: despite his hurt, his grievous deadly wound, he spoke,—he knew full well …
The unperiodic enjambment after line 2724 is followed by an appositive to its b-verse in the a-verse of 2725, which emphasizes the tragic poignancy of the situation. It is noticeable that when the poet leaves the pattern of Beowulf maþelode, / bearn Ecgþeowes, it is for significant emphasis. It is also worth noting that this gesture does not mean that the poet is departing from the traditional, because the departing itself is traditional.
We can conclude, therefore, that the Beowulf poet shows evidence of thriftiness and, in so doing, indicates an affiliation with the oral traditional style of Homer.


[ back ] * Portions of this chapter were presented at a lecture, "Recent Comparative Perspectives on Oral Traditional Poetry," sponsored by the Center for the Study of Comparative Folklore and Mythology, the Folklore and Mythology Program, and the English Department, at the University of California at Los Angeles, February 20, 1986.
[ back ] 1. Watts, 1969; see Chapter 4 at n. 18.
[ back ] 2. Benson, 1966.
[ back ] 3. Niles, 1981, 410.
[ back ] 4. Ibid., 409.
[ back ] 5. Ibid., 410-11.
[ back ] 6. A. Lord, 1991, 147-69.
[ back ] 7. Beowulf contains 3,182 lines; Elene has 1,321; Genesis , 2,936. The total of Elene and Genesis is 4,257 lines.
[ back ] 8. A. Lord, 1991, 151-52, 161.
[ back ] 9. Ibid., 165.
[ back ] 10. Benson, 1966, 340-41.
[ back ] 11. Niles, 1983, 138.
[ back ] 12. Fry, 1967b, 1968c.
[ back ] 13. Fry, 1967b, 204.
[ back ] 14. Fry, 1981, 172-73.
[ back ] 15. Parry, 1971, 13.
[ back ] 16. For the conviction that an expression need not be used more than once by a singer for it to be considered a formula, see Bynum, 1987, 103-6.
[ back ] 17. Fry, 1981, 173.
[ back ] 18. Ibid.
[ back ] 19. Riedinger, 1985. An important article by Fry not mentioned in the preceding survey is Fry, 1968a. It stems from Benson, 1966, and continues the discussion of thrift in Beowulf as set forth in Whallon, 1961, 1965a, and 1965b.
[ back ] 20. Riedinger, 1985, 306 no. 30; Krishna, 1982: "A true formula therefore is not simply a repeated phrase but a phrase that fits into a system or set that exhibits extension and, most important, thrift" (77); Niles, 1983, characterizes a system as "a set of verses of a similar metrical type in which one main verbal element is constant" (126).
[ back ] 21. The abbreviations of Old English texts are those listed in Bessinger with Smith, 1978. See Riedinger, 1985, 294 no. 4. The Old English texts are from the ASPR, and the translations of the passages cited by Riedinger are hers, with slight modifications.
[ back ] 22. Riedinger, 1985, 297.
[ back ] 23. Ibid., 300.
[ back ] 24. See A. Lord, 1960, e.g., 37, 47-48, 297-301. For the importance of "sound-patterning" in the concept of the formula, see Creed, 1981, 19-20.
[ back ] 25. Riedinger, 1985, 303.
[ back ] 26. Ibid., 304.
[ back ] 27. Luethans, 1990.
[ back ] 28. Notopoulos, 1959; Hainsworth, 1968 and 1978; Hoekstra, 1964; Nagy, 1976 [and 1990a, 50-51]; Janko, 1982, II, 19-24, [and 1992; see index s.v. formulae].
[ back ] 29. Riedinger, 1985, 305.
[ back ] 30. Ibid., 310.
[ back ] 31. Whallon, 1961 and 1969.
[ back ] 32. The translations from Beowulf in this section are from Clark Hall, 1950.
[ back ] 33. For more on "Beowulf spoke," see A. Lord, 1991, 147-69.