7. The Ballad: Textual Stability, Variation, and Memorization

It has become apparent that, because the study of oral traditional literature has so concentrated on epic, the application of the oral theory to other kinds of oral poetry may lead to difficulty. The ballad has presented one thorny problem, and short forms in general another, the one because of its stanzaic form, the other because of its comparative brevity. I have treated short lyric poems in Chapter 2. In the case of the ballad, it seems possible, as Bertrand Bronson once suggested, that some ballads were more fluid than others in their texts up to the time when they were written down and published, thus establishing a text such as one had for other literary poems. [1]

A glance at some of the texts of the 198 variants of "Barbara Allen" (Child No. 84) in Branson's The Traditional Tunes of the Child Ballads makes one pause. [2] Bronson vividly summarizes "Bonny Barbara Allen" as follows: "This little song of a spineless lover who gives up the ghost without a struggle, and of his spirited beloved who repents too late, has paradoxically shown a stronger will-to-live than perhaps any other ballad in the canon." [3] The texts of this ballad are not as close to one another as one might have expected them to be from what is often assumed to be a tradition of memorization. [4] There are some close texts, but others are quite different. The idea that a memorized text is behind all of these variants would not be easy to defend. {167|168}

An investigation of the textuality of this ballad is in order. I have chosen to examine the opening stanzas of eight variants from the group beginning with a form of "In Scotland I was born and bred." This group, a relatively small number of Branson's 198 variants, provides a convenient unit for study. [5] The eight versions embrace three of the four classes of tunes described by Bronson for "Barbara Allen." [6] These variants begin:

From a study of these first stanzas it is clear that the unit of composition is mainly the couplet. One cannot predict with certainty from the first couplet in the stanza what the second couplet will be. Five of these eight variants begin with the line "In Scotland I was born and bred"; one reads "In Scotland I was bred and born," and another begins similarly, "In Scotland I was bred aden born." The eighth begins "(In) London I was bred and born." The second line is somewhat less predictable than the first. No. 1 has "In Scotland is my dwelling"; No. 9, "And England is my nation"; No. 12, "In Scotland I was dwelling"; No. 27, "In Scotland was ė my dwelling'"; No. 51, "In Scotland was my dwelling"; No. 94, "O, there it was my dwelling"; No. 127, "In London I was dwelling"; and No. 156, "(In) Scotland was my dwelling', O." In spite of this lack of complete fixity, the couplet holds together pretty well as being something that is varied, usually beginning with "In Scotland" and ending with "dwelling."

The "In Scotland I was born and bred" couplet is usually followed by either "I courted there a pretty maid," which rhymes roughly with "bred," or its variant "And there I loved a pretty maid" or with "A young man on his death-bed lay," which, of course, rhymes with another opening couplet, "'Twas in the merry month of May." The opening couplets "In Scotland I was born and bred" and "(In) London I was bred and born" are also followed by a couplet beginning with "I fell in love with a nice young girl," a variant of "I courted there a pretty maid." {169|170}

Although a singer may have arrived at a version that became habitual with him or her, probably by some form of "remembering" or memorization—but not memorization of a single already fixed text—somewhere along the line, couplets have moved from association with one other couplet to still another. The couplets, moreover, themselves are not stable. Of the eight "In Scotland" couplets, as we have seen, no two are exactly alike, although several are close.

When one proceeds to the second stanza, it becomes apparent that the eight versions differ in the pace of the narrative. Four of the second stanzas begin with some mention of courting. No. 27 has "I courted her for a month or two"; No. 94 has "I courted her in summer time"; Nos. 127 and 156 have "I courted her for seven long years." It is noticeable that in No. 27 and No. 94, the courting has begun already in the third line of the first stanza. The second stanza of No. 12 commences at a different moment in the story: "He sent his servant to her house," whereas in No. 1, Barbara already goes to the young man's bedside. In No. 27 a striking deviation is that the narrator is the servant not the young man himself.

We have thus seen something of the variations in text in the first two stanzas of six of the eight variants of "Barbara Allen"; (two of the eight lack a second stanza). We have observed the relative instability of the text in the traditional ballad and its propensity toward variation. An examination of these eight variants bears out Bronson's remarks on variability in the tunes and texts of the Child Ballads:

Let me now turn to the variants of one of the shorter songs, a ballad, in Serbo-Croatian from the Milman Parry Collection. [8] What can we find out about its textuality? I have taken the beginning of five versions of San usnila Hasanaginica, "Hasanaga's wife dreamed a dream" (Bartók No. 12a, Parry No. 6391a) and compared them for their sense of textuality, or rather for stability of text in a song sung or dictated by women in a very closed group in Gacko, Herzegovina, in 1935. [9] I have numbered them 12a, A, B, C, and D for convenience of reference. Bartók 12a is sung in couplets, as is shown below. The dictated texts have some differences from the sung text purely because they are dictated and not necessarily written down well by the scribes, who were sometimes school children.

Here is a summary of the 130 lines of Bartók-Lord No. 12a:

The first line is common to all the texts, except that the lady is Hasanaginica in some (12a, A, and B) and Alibegovica in others (C and D). The second line, sung together with the first in a couplet, San usnila, u snu se prenula, "She dreamed a dream and while dreaming she started up," repeats {173|174} the first half of the first line and adds a new idea in the second half. This is a common method of line, or couplet, structure, which I have sometimes described as fitting three ideas into four metrical slots by repeating in the second line an idea introduced in the first. Text D has the same second line as 12a, although the lady is Alibegovica, not Hasanaginica. The second line in C, San usnila, pa se probudila, "She dreamed a dream, then awoke," uses a different verb in the second half of the line, with essentially the same meaning, "she woke up." Text B is defective in this line, U snu se prenula, "while dreaming she started up," the scribe not bothering to repeat the first half of the line and writing down only the second half, which is the same as in 12a and D.

In Text A, lines two and three: "intervene" between what are lines one and two in the other texts, with the result that line two of the other texts is line four of Text A, and it (San usnila, u snu se prepala, "She dreamed a dream, in her dreaming she took fright") repeats the first half of line one but has still another verb in the second half. Lines two and three of Text A will be found in another position in the other texts. They are descriptive of the dream, and there are several places in which they might logically occur and, as a matter of fact, do occur.

In 12a, lines three and four go together in a couplet just as one and two did. Bartók No. 12a is the only one of the five texts to use the verb (pro) budi 'awoke' (line three); the others use doziva 'called to'. The person called in 12a, A (line five), B, and D is "daughter Melećhana"; in C it is Ajkunu đevojku 'the maiden Ajkuna'.

The second line of this couplet begins with Ustaj 'arise', in 12a, A, and B, followed by a vocative for the daughter. In D, the vocative is repeated in both halves of the line, Ćeri moja, mila Melećana, "My daughter, dear Melećana." {174|175} Once again, Text C is different (as in Ajkuna đevojku). The text has no vocative but begins immediately with Čudan sam ti noćas san usnila, "Last night I dreamed a strange dream."

The next couplet in 12a is: This couplet is more or less the same in A, but it is amplified to three lines in D: There is a balance in some lines of the poetry between Bog 'God', and ljudi 'people', in the traditional poetry, and that has lengthened the second line of the couplet, thus requiring a third line for its completion. Texts B and C omit the idea that the mother has fallen ill and take up, rather, the idea of her dying only in connection with the interpretation of the dream, which appears later in the other texts.

The next three new lines in 12a should be taken together because they introduce the lady's dream: Text D gives a three-line equivalent to the three lines above, and they are in the same sequence: Text B has also a three-line equivalent to the 12a lines in question, also with slight variations: {175|176} These lines, however, follow immediately on the Ustaj 'arise', line because the mother's illness has been omitted; or rather, there is no mention of it. Text A has the two lines describing the dream but not the line introducing them, but they are placed immediately after the first line of the song, as we saw before, "interrupting" the first and second lines.

Text C does not give a description of the dream and keeps only the line Čudan sam ti noćas san usnila, "I had a strange dream last night," adding Brže će ti majka umrijeti, "Your mother will soon die." Text C telescopes the song, shortening it considerably.

Bartók No. 12a then gives the interpretation of the dream in four lines: Text B has approximately these same lines of interpretation but introduces them with a vocative line: Šćeri moja, Melećhanu, "My daughter, Melećhana." Text D gives only two lines of interpretation, corresponding to the first and the last lines of 12a: Text C has only two of the four lines of interpretation found in 12a, namely: Finally, Text A has only one line of interpretation, and it falls hard on the {176|177} heel of the mother's illness: I tvoj će se oženiti babo, "And your father will marry again."

These sample versions present fewer variations than we saw in the "In Scotland I was born and bred" versions of "Barbara Allen." The South Slavic examples, as noted, are all from the same small district and represent a song that was well known. Even so, they reveal many differences in text. The analysis is sufficient, I believe, to illustrate what I mean by a sense of textuality, that is, a sense that a song has a recognizable text and that the singers recognize that fact. But the existence of such textuality does not by any means imply a fixed text or an attempt at rote memorization. The ballad of "Hasanaga's wife" is "remembered" rather than memorized. [12]

Questions concerning the nature of the transmission of some Anglo-Saxon poems and of Anglo-Scottish ballads arise in connection with the thesis proposed by Alan Jabbour in his article "Memorial Transmission in Old English Poetry." [13] His research has been thorough so far as the Anglo-Saxon texts are concerned, but there is room to reconsider his interpretation of the cause, or causes, of the variations that he has so skillfully and clearly set down. His evidence—other than the Old English texts themselves—for memorial transmission is founded in English and Scottish ballad tradition. There are clearly parts of traditional ballads that are comparatively stable, but also clearly, there are variations, as we have seen in the case of "Barbara Allen." In using the evidence of ballad transmission in attempting an interpretation of Old English poems, parts of which are stable and parts variable, one must remember that the ballads are traditional in subject and style, with a history of transmission over a fair, if varying, number of generations and singers. In contrast, Soul and Body, Daniel, and Azarias are scarcely traditional subjects and, so far as I know, were never sung by many of the populace over generations. To use ballad variation as an analogue to explain variation in some Old English poems seems to me to be methodologically unsound.

We are left, then, with two problems instead of one, neither of which is adequately explained: first, the variations in some Old English poems that cannot be accounted for by scribal error, and second, the processes of transmission or even of composition, or both, of traditional ballads. These are separate problems. I find it easier to approach that of traditional ballads, where material for analysis is abundant and more varied than that of the comparatively small group of Old English poems involved, which would have been known to a restricted number of persons. I might add that one {177|178} need not confine one's observations to British, Scottish, or American ballads but should include Danish, Swedish, German, or any other ballads of similar form and tradition, whatever the language or culture may be.

Jabbour rejects comparative studies based on "modern foreign tradition." He says, "The aim of the foregoing discussion has been to disengage the speculations about Old English oral tradition from a point of view shaped by the study of a quite different tradition, and to suggest that, in the absence of first-hand evidence about Anglo-Saxon oral practices, the analogy of later British oral tradition should carry greater weight than the analogy of modern foreign traditions." [14] Yet, while having objected to a comparison between Beowulf and South Slavic epic tradition, he accepts for analogy with certain Old English religious poems the study of the far different tradition of ballad, albeit in the same geographic frame. Jabbour's, and others', primary difficulty is in not understanding the difference between a fixed and a fluid text, and in the case of ballad and epic, between stanzaic and stichic form.

In speaking of the English ballad, one must be careful. What ballads is Jabbour talking about? They are not all alike in fixity of text. If Jabbour wishes for a "memorial tradition" in English with which to compare Soul and Body or Daniel or Azarias, he might look to the fixed text of the nontraditional broadsides. The broadside sometimes starts with a fixed text composed by someone, whose name is known. [15] We know its first form, and knowing that, we can note divergences from it. That is memorial transmission of a fixed text, which is not an oral traditional process.

The solid "core" of traditional English ballad, the ballad tradition itself, is a different matter entirely. Here the case for memorial transmission is not open and shut; for the text of a ballad is not fixed, not even after it has been published, less so, of course, before. One is dealing with lines or couplets that may be used in more than one stanza and in more than one ballad, provided the meter and tune allow. The exact wording of such lines and couplets is not predictable, but they maintain a high degree of stability. The topoi of the ballads are too well known to need extended comment. Such lines as "She mounted her milk-white steed," "he on his dapple gray," and the like are useful with little change in a number of songs. These lines and couplets are like the lines and couplets of the formulaic style. On this level, the process of formation of stanza lines and probably their later re-creation seem at some {178|179} time to have been much like formulaic composition. Here we have, or have had, not memorial but "improvisational" transmission, that is, composition in performance.

It is paradoxical that, on the one hand, in the ballads we find oral traditional texts displaying a kind of textual stability sometimes characteristic of nontraditional "written" processes; on the other hand, in the nontraditional written texts of Old English Soul and Body and parts of Daniel and in Azarias, we find textual variation characteristic of oral traditional texts. If we remember that Old English written Christian poetry, even that translated from Latin, uses a formulaic technique of composition derived from oral traditional poetry, then we can conclude that Old English written Christian poetry derived its technique of variation in reproduction of text as well as its composition from oral traditional poetry. On reflection, it is not surprising, because reproduction of text is a kind of recomposition.

It is not my purpose here to analyze fully balladic formation and structure. Suffice it to say that to call it as a whole "memorial" is simplistic, because that implies fixed texts, and such are not to be found, except possibly at this latest hour when all singings stem from the written, fixed texts of the collections. What is more for our present purposes, to use balladic formation and structure as an analogue to explain the variations in Old English Soul and Body, Daniel, and Azarias is misguided. The two groups of poems are quite different in all ways, even, I admit, pace Jabbour, in the reasons for variations of text.

Jabbour has done a real service, however, in calling attention to the multiple manuscripts of Old English poetry; for they are often ignored and his isolating of those poems in which variation is clearly not fully accounted for by scribal error is very useful. Still, I am not at all sure of the reason for the variations.

At the time when Jabbour was working on this subject for his dissertation at Duke University, Alison Jones at Canberra was also investigating the variations that occur between the "Prayer of Azariah" as given in the Old English Azarias and another version of it in the Old English Daniel. [16] This research was published and Jabbour referred to it in his own article. Jabbour and Jones, stressing the importance of memory, can account for variation only by lapsus memoriae, which does not by any means explain all the variants. Jabbour did not refer to a later article by the same scholar on the variations in Soul and Body, I and II. [17] {179|180}

In her 1966 article on Daniel and Azarias, Jones examined in detail the variations in the corresponding passages of those two poems: As Jones says, "The relevant sections of the poems are Azarias 1-75 and Daniel 279-364, where the two texts are roughly parallel, and Azarias 76-175 and Daniel 365-415, where some similarities of phrasing occur, but where the two poems go more their own way, which can be seen even from a comparison of the number of lines which the two poems take to deal with the same material." [19] In 1969 she said of Soul and Body: She concludes:

Earlier Jones spoke of oral transmission as "a means of transmission which previous critics have noted to be facilitated by the formulaic nature of the verse, a characteristic revealed in the study of both of these poems." [22] Thus she notes the formulaic character of both Daniel and Azarias, and at the same time, she repeats the fallacy that formulaic verse is easier "to transmit" than nonformulaic. Later she reiterates, "Since it is the formulaic nature of Old English poetry that makes it so suitable for oral recitation, let us look first at some of the variations in formulaic phrases between the two poems." [23]

Formulas and the formulaic style are aids in composition, or were so in their inception, and not "intended to facilitate recitation or memorization," nor do they. If formulas made memorization easier, why are there so many formula substitutions among the variants? Indeed, they make word-for-word memorization more difficult because it is not easy to remember which of two or more similar formulas is used at any given place in a fixed original. The formulas are, I hazard to say, unnecessary and awkward in a purely "memorial" tradition.

In this regard, I was interested to read of a computer experiment attempting to determine the usefulness of rhyming words as a mnemonic device in poetry. Suzanne Petersen wrote: And finally, after other tests, which indicated that only the infinitive showed stability in rhyming position, Petersen concludes:

Petersen's evidence is corroborated by an incident recorded by Milman Parry in his unpublished journal "Ćor Huso" under the date of December 2, 1934. Parry tells of a poem dealing with the assassination of Archduke Franz Ferdinand, which he was reading into the dictaphone. He says, "The poem is written as most of the new poems are, in rhyme, and while the themes and phrases of the poetry show on the one hand a habitude of the old poetry, other verses clearly indicate the influence of the newspapers." Parry describes how, as he was reading the verses of Gavrilo Princip's speech before he fires the fatal shot, his assistant and scribe, Nikola Vujnović, himself a singer, recited the line that follows them. Parry says, "When I asked him to recite the poem from the beginning he was able to do so only for twelve verses, and then lost himself in making the rhymes … His explanation which, in its way, is doubtless true, was that it is very easy to forget poems that are rhymed." This incident as well as the data from the computer studies cast doubt on theories urging the "memorial transmission" of some traditional ballads.

To return to Old English, it is to be noted, although Jones does not actually say so, that Azarias 76-175 and Daniel 365-415 are so far apart, in spite of occasional similar phrases such as sunne ond mona (Azarias 77) and sunna and mona (Daniel 369) that there can be no real question of "memorial transmission." [26] {182|183}

Lord's text of his chapter ends here. One is left, therefore, with serious doubts, pace Jabbour, of the "memorial transmission" both of the Anglo-Scottish traditional ballads and of the Old English Soul and Body, Daniel, and Azarias. With regard to the Anglo-Scottish ballads, the word memorization is deceptively enticing in a truly traditional setting. Lord liked to maintain a distinction between what the singer "remembers" by a natural and informal mental process and what he or she is alleged to have reproduced by memorization. The very process of oral composition makes an attempt at close or exact memorization unnecessary. First and foremost, in traditional poetry there is no fixed text to memorize. An "original" or archetypal text is the chimerical goal of literary critics. Behind Jabbour's memorial transmission there lies the assumption of an "original" text. He says, "If we discover … that in a certain tradition the variants of a song show a history of word-for-word or phrase-by-phrase oral transmission from a known or presumed archetype, we may describe the tradition as memorial." [27] For the Child Ballads accepted into Toelken's "oral canon," [28] such a fixed text was nonexistent, as is the case with all truly traditional lore, with the exception of very short forms.

As for the religious Old English poems discussed by Jabbour (but not for all Old English poetry, notably Beowulf), Lord used the words transitional or perhaps mixed to describe them. [29] He modified his views on the "transitional text" as expressed in The Singer of Tales in later articles, especially in "The Merging of Two Worlds: Oral and Written Poetry as Carriers of Ancient Values" and in "The Nature of Oral Poetry." [30] The introduction of writing (and a new literary style) did not come, as Jabbour maintains, "at a stroke." For a considerable time, established formulaic habits continued into a period of style known to have been influenced by written literature.

Jabbour has a different view of the transitional text. Influenced by Lord's 1960 position that there is no transitional stage between oral and written style, Jabbour says:

Katherine O'Brien O'Keeffe takes issue with Jabbour's advocacy of the memorial transmission of Old English poetry. [32] As I explained in the Addendum to Chapter 4, O'Keeffe by using very refined paleographic methods, plots for Anglo-Saxon manuscripts and their scribes the stages on the continuum from orality to literacy. Her voice adds support to Lord's position in regard to the transitional milieu of Old English religious poetry, even though she concentrates not on the poet but on his reader or scribe.

In regard to both the ballad and Old English poetry, we must return to the search for what is traditional and to methods of measuring the degree and the kind of traditional devices that the poetry shares. We need an understanding of a style that has remained fluid, subject to variation by a compositional process rather than by lapsus memoriae or by mere scribal error or incompetence.

In connection with Jabbour's thesis of memorization, the more recent work of Murray McGillivray comes to mind. [33] His carefully measured attempt to demonstrate the transmission by memorization of four Middle English romances is not entirely convincing. The parallel passages he cites for comparison among different versions of the same romance are anything but fixed in their text. The manuscripts of the romances, by McGillivray's admission, show "massive variation from text to text." Groups of two or three lines, moreover, which are found to be transposed and to be inserted at different points in the narrative, rather than proving faulty memorization of set passages, could possibly illustrate blocks of lines or formula clusters that are useful in composing a text that is essentially fluid in nature. The author's inability to define an archetypal text for a group of manuscripts of the same romance is ominous for a theory of memorization. An "original" version that was available for the minstrel to memorize, well or badly, fades into the {184|185} distance to the vanishing point. One is faced with a group of texts the variations in which are not fully explained.

Relevant to the problems treated by Lord in the present chapter on the ballad is Buchan's discussion of three stages in the composition and transmission of ballads in the northeast of Scotland. He defines the first stage as a period of oral composition, the second a transitional period, and the third and final stage as one of memorization when habits of literacy have taken complete hold. The oral period he describes as a time when "the Northeast was still largely nonliterate." The concept of a fixed text was not yet established, but ballad singers were creative, or at least "re-creative." They were not bound by the words of a ballad as they had heard it but maintained something closer to stability of narrative rather than to stability of text, freely making changes at will. [34] Buchan's choice of Mrs. Brown of Falkland, who learned her ballads mostly before 1789, as the exemplar of the oral period of northeast balladry has not met with the approval of several scholars; for she was an educated person even though she learned her songs in her youth from persons of the old oral tradition and herself carried on that tradition. [35]

The example of transitional texts provided by Buchan are the ballads of James Nichol of Strichen, who died in 1840. Buchan traces the tendencies that emerge in the ballads of the transitional period, many of them traits that stem from changing social conditions. He finds the transitional method of singing "re-creative but rather haphazardly so." In Nichol's ballads, "sometimes a sequence of balances is merely a string of paired stanzas rather than an organic scene whose units are intricately related." [36] What Buchan calls "the annular device," that is, ring composition, an artistic mark of nonliterate style, is found rather infrequently in Nichol's texts.

Buchan notes that "English, the language of literacy for Scots, made ever-increasing inroads into the old formulaic style. Everywhere in the texts we can see the results of the new education: in words like 'espied' … and in lines such as 'Lay gasping on the ground'" (231). The influence of the broadside is {185|186} evident in beginning lines with such detail as "Upon the eighteenth day of June." Especially telling in the ballads that Buchan calls transitional is the "religiosity that now incongruously appears in the ballad texts [which] may be traced to the spread of devotional literature. Evangelical terminology accords ill with the ballad context" (241).

The modern period of northeast ballad tradition, according to Buchan, began about 1830. With the introduction of the printed ballad texts into the tradition comes the memorization of the ballads, the third stage of transmission. For his exemplar of this phase Buchan chose the ballads of Bell Robertson, who was also a writer of pious and spiritual poems. Among the several signs of the memorization of Robertson's ballads is their incompleteness. He reports that only one-third of the eighty-four ballad pieces in her repertory could be called complete. Because she recited rather than sang her ballads, the lack of music to aid her memory may have helped to explain the fragmentary nature of her pieces. She often did not understand particular words in her text or believed that they could have been corrupt, but she insisted on giving what she had heard. "Not only was it a case of her memorizing texts sung by other singers but also of her learning by rote off the printed page" (251).

As we approach the end of the twentieth century, when ballad singing remains a continuing aspect of popular culture, whether performed before live audiences or brought to our ears by record albums or compact discs, it is helpful to look back to the tradition of the Scottish northeast. Of special importance for this volume is the information that the Anglo-Scottish tradition teaches us about comparative stability of text and matters of transmission and memorization, and it helps, moreover, to define when the term memorization is appropriate for the ballad.