The Quantitative Paradigm: George Hickes’s Thesaurus (1703–1705)

Anglo-Saxon scribes wrote their vernacular poetry in continuous format, without line breaks, a presentation regime that created basic challenges for early scholars, and continues, on occasion, to do so today. The first modern scholar to recognize Old English poetry as poetry was probably Francis Junius, who, in 1650, undertook to edit the biblical paraphrases contained in the manuscript that now bears the shelfmark Junius 11 in the Bodleian Library.⁸ Junius recognized that he was dealing with poetry. In this, he was aided by his manuscript source, which is distinguished among the major codices of Old English poetry by its uniquely persistent metrical punctuation.⁹ The scribe wrote in continuous format, as always, but he frequently entered a raised point between half-lines, thus lending visual perspicuity to the basic unit of metrical composition. Junius extended this markup regime in his 1655 edition, adding some points omitted by the scribe. He similarly marked off half-line units in his transcription of poems in the prosimetrical version of the Old English Consolation of Philosophy; these transcriptions formed the basis for Christopher Rawlinson’s 1698 edition, where, for the first time, Old English poetry was lineated.¹⁰ Individual verses (that is, what we now recognize as half-line units) were set out by Rawlinson in columns, each on their own line.

Once these texts were recognized as verse, the question naturally arose: verse of what kind? At the turn of the eighteenth century, George Hickes affirmed that the earliest English poetry was quantitative, constructed on the same principles as classical Greek and Latin verse:

In Hickes’s estimation, all that lay between modern scholars and a complete knowledge of Old English meter was the identification of vowel quantities. No doubt, this position was motivated in part by extraneous and unanalyzed commitments: admiration of the Greek and Latin poetry on one hand, and undisguised distaste for modern English verse on the other. The chapter “De Poetica Anglo-Saxonum” is sprinkled with deprecations aimed at seventeenth-century English poetry, passages in which Hickes identified himself as a kindred spirit to the Elizabethan quantitative reformers.¹² Yet, unlike that earlier generation of classicizers, Hickes held out no hope that quantities could be made a principle of modern English versification. On the contrary, his comparisons between modern and ancient English were in each case intended to show that the very features that distinguish Old English from the contemporary language also align it with the classical languages known to have quantitative systems of versification.

Hickes was more dependent than Junius had been on graphic signaling of verse boundaries: where metrical punctuation was absent, he often failed to recognize a metrical boundary. Nevertheless, this much was clear: whereas modern English poetry tended to have lines with equal numbers of syllables, the verse units identified by Junius were syllabically unequal. Even if the syllabic values of double vowels and final -e were unknown, Hickes reasoned, those two sources of uncertainty could account for only a fraction of the observed variation. In the classical Greek and Latin verse, variations in syllable count were a function of a line’s foot-structure (for example, the alternation of hexameter and pentameter lines in Latin elegiac verse) and, second, of metrical resolution (that is, the possibility of putting two short syllables in the place of a long one). Hickes reasoned that the same factors must be responsible for observed variation in Old English.

To this argument from syllable count, Hickes adduced three other inferential arguments that Old English poetry was probably quantitative. The Old English verse lacked end-rhyme; its syntax was characterized by bold inversions of word order; and its lexis, Hickes affirmed, was distinctively polysyllabic. By each of these criteria, Old English poetry grouped with the quantitative poetry of classical Greek and Latin, as against modern English poetry. Hickes also ventured a small number of scansions. He proposed to scan certain four- and six-syllable verses as spondaic. The following verse is an example. For the sake of fidelity to the state of knowledge in this early scholarship, I omit the macrons now used to distinguish long vowels; they will be reintroduced later in my exposition.

Hickes did not indicate vowel lengths typographically and did not explain his scansions. Nevertheless, a tentative reconstruction of his reasoning is possible. Since he classified (1) as spondaic, he presumably assigned a long vowel to Þa (correctly, as it happens). According to Latin quantitative prosody, the third syllable of the line would be long by position: his is followed by an initial consonant. If Hickes believed Old English h had the prosodic function of other consonants, then the second syllable of the line could also be scanned long by position. This would be a departure from Latin prosody, but Hickes had earlier conjectured that h might have been pronounced harshly.¹⁴ The result is three spondees.

A few pages after his list of spondaic lines, Hickes supplied another short list, this time of lines that he proposed to scan as adonics (that is, a dactyl followed by a spondee).¹⁵ An example is:

Here Hickes must have assumed a long vowel in the first syllable of hæleða. We now know that to be incorrect. Regarding the length of the two subsequent vowels in this word, Hickes guessed correctly; they are short. Examples (1) and (2) will serve as recurring points of reference in my exposition below. For now, the important point is that Hickes recognized that such scansions were beyond his or anyone else’s present knowledge of the Anglo-Saxon language. Accordingly, the task he set himself was not to precipitate unknowns into knowns, but instead to gather and present evidence that the unknowns really existed as such – that is, that quantity was present and functional in the poetry. He urged that observed irregularities in syllable count would one day be recognizable as the consequence of as-yet undiscovered quantitative patterns.

It was a reasoned hope. Yet it is also possible to hear a note of desperation, particularly in Hickes’s repeated comparisons of Old English verse to the Pindaric. Pindar’s odes served as a model of lyric exuberance in English poetry at the end of the seventeenth century, but his exceptionally complex metrical patterning was not yet understood.¹⁶ This fact bears significantly on Hickes’s claims for Old English poetry. The references to Pindar in the chapter “De Poetica Anglo-Saxonum” stand as a tacit acknowledgment that, if Old English poetry really had been quantitative, then it must have been composed of metra at least as complex and various as any such produced by Greco-Roman antiquity. Indeed, comparison to Pindar was almost the only way that Hickes could lend credibility to his claim that Old English verse was, despite its bewildering variety, also the expression of a metrical system.

The Paradigm Shift, 1765–1868

The Thesaurus was published between 1703 and 1705. By the second half of the eighteenth century, Hickes’s metrical classicism could be ignored. Then, at the beginning of the nineteenth century, the quantitative theory was swept definitively aside: reference to quantities was replaced by reference to “emphasis.”

The concept was not new. George Gascoigne had set out rules for the patterning of “emphasis,” or accent, in his Certayne Notes of Instruction Concerning the Making of Verse or Rhyme in English (1575).¹⁷ In the eighteenth century, “emphasis” had become a key concept in discussion of English prosody, as a search in the Princeton Prosody Archive demonstrates. The concept had not, however, previously been extended to the vernacular poetry of the Anglo-Saxons. As such, it required definition from the ground up. “Emphasis,” wrote Joseph Bosworth in his Elements of Anglo-Saxon Grammar (1823), “is a perceptible stress of the voice laid upon a syllable, or a word.”¹⁸ Bosworth identified syllabic emphasis as “the superior energy with which at least, one syllable of a word is enunciated,” and he explained that the emphatic syllables must be separated by “remiss or feeble” ones in speech, for “[s]everal emphatic syllables cannot be conveniently enunciated in succession.” There is a tone of bold linguistic generalization in these pages. “It appears,” Bosworth wrote, “that in language emphasis and remission occur at certain intervals. On these depends rhythm, the vital principle both of speech and song.” In a footnote, he affirmed the conceptual break with classical frameworks: “The Greeks and Romans regulated their verse by the length of syllables … But the Anglo-Saxons modelled their verse by rhythm or metrical cadence.” Or, again: “in Saxon and in all modern languages of Gothic origin, [emphasis] holds the place of the Roman and Greek quantity.”

In these sentences and in the account of Old English poetry that they introduce, Bosworth was communicating the newest philological scholarship. His most important sources were less than a decade old. John Grant’s 1813 Grammar of the English Language had described the operations of emphasis and rhythm in spoken language.¹⁹ In the same year, John Josias Conybeare had proposed that Old English poetry was organized not by quantity but “emphasis”: verses of Old English, Conybeare ventured, “will be found to consist for the most part of feet of two or three syllables each, having the emphasis on the first.”²⁰ Conybeare would presumably have scanned (2) hæleða waldend as a three-syllable foot followed by a foot of two syllables (Sxx|Sx).

The great Danish scholar Rasmus Rask, who was another of Bosworth’s authorities, had gone several steps further in his 1817 grammar of Old English.²¹ Rask had understood a fundamental point: the alliteration that mattered in early Germanic verse – the alliteration described in Snorri Sturluson’s thirteenth-century ars poetica – was located at the onset of stressed syllables.²² By seeing that alliteration and stress were linked, Rask was able to bring the Old English poetry to bear on the question of stress-placement, and – in a second move – bring this sharpened understanding of linguistic stress to bear on the question of Old English meter. Subsequent studies would give many more iterations to this feedback loop between study of language and study of meter. For Rask, the result of his linguistic inferences was twofold. He concluded that Old English verse had precisely two emphatic syllables per verse; he also made a rough division between the types of linguistic material that receive emphasis and those that do not. Where Conybeare had proposed to scan

with three feet and perfectly alternating rhythm (Sx|Sx|Sx), Rask objected that emphasis would fall only on the root syllables of major words.²³ He therefore scanned the verse Sxxx|Sx. “[F]or those who wish not to compose A.S. verses, but merely to analyse such as they meet with, it is easy to determine the metre,” he affirmed; “The chief syllable in each word bears the accent. Compound words, consisting of two independent and, in themselves, significant words, are accented on the first.”²⁴ In scholarship of the early nineteenth century one can sense the sparkle of discovery – of recognition that one has made a break into a new way of formulating a problem.

Critical notices of Hickes were exemplary of this new spirit. Rask, the pioneering comparatist, scolded Hickes for having “possessed so little of the spirit of discovery.”²⁵ Conybeare stated that the great and learned author of the Thesaurus “appears perhaps no where to so little advantage as in the pages which he has dedicated to this topic [that is, meter].”²⁶ Yet, in estimating their accomplishment, these scholars looked as much to the intervening era as to Hickes himself, and their progress stood out sharply. During the second half of the eighteenth century, as Hickes’s classicism lost support, Old English poetry had come to appear hopelessly disorganized. In an essay accompanying his 1775 edition of Chaucer, Thomas Tyrwhitt had claimed to discern nothing resembling metrical organization in English prior to the Norman conquest.²⁷ Tyrwhitt averred that the English owed meter, as such, to their Norman conquerors. George Ellis and Sharon Turner agreed. Ellis reported in 1801 that the “mechanism and scheme of versification [of Saxon poetry], notwithstanding all the pains which Hickes has employed in attempting to investigate them, are still completely inexplicable.”²⁸ Turner, meanwhile, commended the unshaped roughness of Anglo-Saxon effusions. In this “rude and barbaric state” of English poetry, he discerned a true expression of the culture’s primitive energy.²⁹

During this same interval, early Germanic poetry was christened “alliterative.” The historical irony should be apparent: early Germanic poetry acquired its Latinate denomination during precisely the interval when the Latin sciences of language were, in all other respects, losing their authority to expound it. Despite their opposition in all other respects, the outgoing quantitative paradigm and the emergent accentual one would at least agree that Thomas Percy’s “alliterative species of versification” designated a metrical system by a linguistic phenomenon that could not per se constitute or organize a meter. Hickes, one of Percy’s sources, had been clear about this: he recognized the alliteration in Old English poetry and considered it important, but as a rhetorical ornament, not an element of meter. He treated alliteration at the head of his section on “incidentals,” where it is followed by rhyme, metaphor, metonymy, apposition, and so forth, down the line of classical figures and tropes.³⁰ Conybeare agreed. In the same paper in which he rejected Hickes’s quantitative theory, Conybeare stated that “alliteration (which indeed requires but a short notice) will be more conveniently treated of after we shall have ascertained the existence and nature of that metre of which it forms the chief ornament.”³¹ The importance accorded to alliteration in Percy’s “Essay on the Metre of Pierce the Plowman’s Visions” was therefore an aberration. It emerged during an interval when scholarship lacked a sure approach to early English poetry: the classical sciences of language had already lost their authority, yet the first articulations of a new paradigm still lay several decades in the future. Tyrwhitt’s confession that he could find no metrical organization in Old English poetry was one symptom of an epistemic interregnum. Percy’s christening of it as “alliterative” was another.

For Conybeare, who looked back on this period from the standpoint of new conviction, the position that most needed answering was not Percy’s elevation of alliteration but Tyrwhitt’s skepticism. The unsupportable classicizing “extravagance” of Hickes had driven the esteemed editor of Chaucer into “the opposite extreme.”³² Between the one position and the other there was an unfortunate symmetry. The concept of “emphasis” or “accent” broke this binary deadlock by projecting a third position beyond both the thesis of classical quantities and the absolute skepticism engendered by it.³³ Tyrwhitt had claimed he was “unable to discover any material distinction of the Saxon Poetry from Prose, except a greater pomp of diction, and a more stately kind of march.”³⁴ That “more stately kind of march” would now be recognized as accentual rhythm. What had previously appeared as no principle at all was now apprehended as the expression of a different kind of metrical system – or, rather, as the same kind of metrical system employed in all later English poetry.

Conybeare had brought the oldest English poetry into agreement with understandings of the modern language and its prosody. Edwin Guest’s genius was to perceive this new epistemic configuration and give it a material body. His two-volume History of English Rhythms (1838) made accentual rhythm the uniform organizing principle of English poetry, from Cædmon to the present.³⁵ It remained for Walter Skeat to consolidate the previous half-century’s advances by supplying the new research paradigm with its own terminology:

Skeat had recognized a signal ambiguity in previous scholarship. Apart from Guest, each of the pioneers of the accentual paradigm had retained the classical names of feet, even as they broke with the classical schema to which these names belonged. Skeat hoped to clear this residual confusion away once and for all. Maintaining that “the whole terminology of English prosody, if it is not to be misleading and fruitful in all kinds of errors, has yet to be invented,” he set himself the task of terminological invention.³⁷ He proposed that one should speak not of long and short syllables, but loud and soft ones. He retained the concept of a metrical foot, but elaborated a new terminology in which the name of each foot instanced the stress contour it named. A single loud syllable would be called a “tone”; a loud followed by a single soft would be called a “tonic”; a loud followed by two soft syllables a “dominant”; and so forth.³⁸ Skeat’s musicological terminology should not obscure his accomplishment, which was to crystallize a sequence of metrical theory begun 60 years previously. Whenever later prosodists have described English alliterative verse as “accentual” or “strong stress,” they simply report the state of knowledge encapsulated in this essay of 1868.³⁹

Elaborations and Challenges

The scansions adduced by Conybeare, Bosworth, Rask, Guest, and Skeat have long since ceased to be relevant. Yet the paradigm established by these scholars between 1814 and 1868 has a history that extends to the present. The most important features of the subsequent history may all be traced to the separation of medieval English into “Old” and “Middle.”⁴⁰ For Skeat, the pre- and post-Conquest poetries were analyzable in precisely the same terms, composed of the same types of feet and organized by the same accentual principles. Soon afterwards, this unity would begin to pull apart; the accentual paradigm would develop along divergent trajectories in Old and Middle English metrics.

Of these two trajectories, the one in Middle English metrics has proceeded more haltingly, but with fewer complications. At the end of the nineteenth century, Karl Luick described the accentual patterning of the late Middle English alliterative b-verse almost exactly as Thomas Cable and Hoyt N. Duggan would 90 years later. Fourteenth-century alliterative poets counted syllables in the second half of the line. There are two basic patterns:

• type 1    (x)Sx…xSx

• type 2    x…xS(x)Sx

These two b-verse patterns are a central component of all recent descriptions of Middle English alliterative verse, including the one I made in the Introduction to this book. We can now place them within a history of thought about English meter. No less than Skeat’s “Essay on Alliterative Poetry,” the Cable–Duggan b-verse patterns grasp English meter as a function of the contrast between stress and unstress; the difference is a more precise template for arrangement of unstressed syllables, coupled with a rejection of Skeat’s metrical feet. With the work of Cable, Duggan, and subsequent researchers, we see a long-delayed elaboration of the accentual paradigm in Middle English alliterative metrics. Several factors, to be discussed at the end of this chapter, prevent one from characterizing this victory as absolute.

In the field of Old English meter, the accentual paradigm has been more steadily productive, sustaining continuous research activity throughout the past two centuries, but also more persistently compromised. Irregularities and extraneous factors have proven unamenable to reduction; those irregularities present from the outset have been joined by others that have emerged only gradually. A history of this research activity would note the work of Karl Lachmann and his school in the mid nineteenth century, as Kühnel does, and the challenges posed to Sieversian metrics in the twentieth century by Andreas Heusler and John C. Pope. Rather than provide a chronological account, I simply sketch three salient problems in the accentual paradigm.

The first is clashing stress. A basic premise of the accentual paradigm is that stressed syllables should either have differentiated prominence or be separated by an unstressed element. Nevertheless, scholarship of the nineteenth century quietly admitted the existence of verses that violate this rule. Conybeare scanned

with stress on tir and the initial syllable of welgade.⁴¹ The chief problem, as Conybeare saw it, was that this verse and others like it has just one complete foot, preceded by what he called a “syllable extraordinary.” He proposed that the “emphasis [of the syllable extraordinary] might be so strongly marked, as to render it equivalent to two,” and Bosworth followed him in this.⁴² As the accentual system of Old English came to be better understood, scholars inevitably became more sharply conscious of clashing stress. Clashing stress followed, as a matter of course, from Skeat’s proposal to recognize stressed monosyllables (“tones”) in his system of scansion. At the end of the century, Sievers would classify verses like the one above as Type D, scanning S|Ssx, with secondary stress on the medial syllable of welgade.⁴³

The Five-Type System formalized clashing stress as a central and prominent feature of Old English meter. Clashing stress has retained this status ever since, though not without provoking nagging doubts and ingenious efforts at obviation. Bosworth had been confident that stressed syllables must be separated. On the basis of vastly more sophisticated demonstrations, twentieth-century intonational phonology likewise affirms that the English language has a powerful preference for alternating accentual rhythm. “When two fully stressed syllables occur in succession,” Cable writes, “the rhythmical pattern runs counter to an idealized norm.”⁴⁴ In Cable’s account, clashing stress should be understood as an artifact of the Sieversian abstraction, not a feature of the meter itself: the task of metrical theory is then to develop a more accurate abstract representation of the meter, one in which stress clash does not appear as such. To this end, Cable has proposed that Old English meter differentiated between four levels of stress and that it included compensatory temporal spacing – or pauses – as a metrical feature.⁴⁵ These theoretical proposals respond to and thus flag a deep problem in the conceptual architecture of Old English metrics: stress clash is at once a result of research within the accentual paradigm and a violation of its most basic presuppositions. The problem, it should be noted, exists in Middle English alliterative verse as well. A solution can only emerge after considering the other two challenges within Old English metrics.

The second challenge is this: quantity never fully disappeared. From the time of Jacob Grimm’s reconstructions of the phonology of medieval Germanic languages, it was evident that vowels in open syllables bearing primary word-stress were not invariably long.⁴⁶ In the 1830s, John Mitchell Kemble championed Grimm’s reconstructions of vowel length, against the vociferous objections of the English Saxonists.⁴⁷ The discrimination between short and long vowels had an immediate value in the fields of etymology and lexicography, where vocalic quantity served to distinguish words identical in spelling. Yet it was not immediately clear how, or whether, the reconstructed vowel system of medieval Germanic languages was relevant to verse design. Grimm himself had expressed doubt on this question. Skeat’s subsequent efforts to banish quantitative terminology were founded in the belief that quantity was entirely irrelevant to early English meter. Nevertheless, even prior to Skeat’s essay, the metrical effects of syllabic length were being noticed and described in Old High German and Old Norse.⁴⁸ These early Germanic meters appeared to treat a short stressed syllable plus the following syllable as a unit equivalent to a long stressed syllable. The phenomenon was termed “resolution.” By the end of the nineteenth century, resolution would be identified in Old English verse as well: it was an integral component of Sievers’s system.⁴⁹ So presented, metrical resolution in early Germanic verse differs sharply from the phenomenon of the same name in Greek and Latin quantitative verse, at least as presented in handbooks.⁵⁰ Nor does resolution accord easily with stress-based conceptions of Old English meter.⁵¹ Skeat would have scanned (2) hæleða as a dominant (Sxx), which may be a good representation of stress contour. By the end of the nineteenth century, however, the root vowel was recognized to be short and the word was usually scanned with resolved stress, assigning the first two syllables to a single metrical position. It was clear that Old English poetry did not owe its metrical organization to accentual contour alone. Quantity had returned, forming a second irregularity within the accentual paradigm.

The third irregularity is the metrical behavior of unstressed prefixes. Although this problem has been recognized only in the last several decades, its slow emergence may be traced across almost the entire history of modern metrical study, under the various names Maalfylding, complement, Auftakt, and anacrusis. We have seen that Rask understood verse rhythm to begin with the first stressed syllable; all syllables prior to the first stress could then be understood, without great difficulty, as a sort of extrametrical prologue.⁵² In a verse like (1) Þā man his riht tōbræc, Rask would have seen an extrametrical prologue of three syllables; the verse proper would begin with riht and would scan SxS. Within Sievers’s system, most verse-initial weak syllables were redefined as integral components of verse rhythm. Sievers Types B and C each begin with one or more unstressed syllables; example (1) is a Sievers Type B verse with a three-syllable initial dip (xxxSxS). Nevertheless, Sievers’s classificatory system left a residue of initial unstressed syllables that could not be accounted for in this way. For such syllables, Sievers retained both the name Auftakt and the rationalization that this linguistic material fell outside the metered portion of the utterance. This feature is illustrated by verses such as those below. I use “)” to set off syllables deemed anacrustic; the sequence “Sr” indicates resolved stress:

Each of these verses belongs to Sievers Type A. Each is understood in Sieversian metrics as having an anacrustic syllable. This was where research remained until it was noticed, first, that the syllables in anacrusis tended to be unstressed prefixes or the negative proclitic ne and, second, that these morphemes exhibit similar behavior interior to the verse.⁵³ This second point is perhaps illustrated most readily by verses traditionally classified as Type A with anacrusis, but which – unlike (5–7) above – have neither an unstressed prefix nor ne in initial position. An example is:

If swā is regarded as anacrustic, this verse forms an exception to the morphological pattern illustrated by (5–7). Example (8) also contains an unstressed prefix, however: be-, between the two stresses. If the rule that operates in (5–7) is not “weak syllables preceding the first stress may optionally be excluded from the metrical count” but instead “unstressed prefixes (and ne) may optionally be excluded from the metrical count,” then (8) may be reclassified as Sievers Type C without clashing stress.

Termed the “prefix license” by Yakovlev, this metrical feature is probably the greatest challenge to the accentual paradigm. “It is very hard,” Yakovlev observes, “to understand how a syllable may be completely omitted from the metrical count [in an accentual meter]: the prominence of the syllable may be low, but it will still be part of the intonational contour.”⁵⁴ Yakovlev’s 2008 thesis may mark the point at which the accentual paradigm will have tipped over into another way of comprehending Old English meter. He calls the meter not “accentual” but “morphological.”⁵⁵

A Second Paradigm Shift: Yakovlev’s Morphological Theory

Stress clash, resolution, and the prefix license have the status of conceptual embarrassments for the accentual paradigm of English metrics: they are smudges in the picture, elements irreconcilable to the structure from which they nevertheless cannot be removed. Yakovlev’s theory does not remove the smudges; instead, it projects them into another conceptual space. The phenomena that always resisted rationalization within stress-based conceptions of Old English meter now reveal themselves as elements of another system.

According to Yakovlev, the standard verse form of Beowulf owed its metricality neither to a fixed count of “major” stresses, nor to a minimal set of recurring accentual contours, but instead to a relatively simple count of four metrical positions per verse. Given this point of departure, the success or failure of Yakovlev’s theory rests largely on its ability to define a metrical position and show how the words and syllables that make up actual verses of Old English map onto a basic four-position frame. For this task he is able to draw on a deep file of prior scholarship. For, if his theory is directly contrary to the general assumption among metrists since before the time of Sievers, it is also a direct development of the single most important contribution to this modern tradition, that of Sievers himself. Here we must take a closer look at Sievers’s accomplishment. He demonstrated that most verses of Old English conform to one of five basic rhythmical patterns. Since their original presentation, the Five Types have always been illustrated by their simplest realizations (examples are from Beowulf):

1. A Sx|Sx: gomban gyldan, “to pay tribute” (11a)

2. B xS|xS: hīe wyrd forswēop, “events swept them” (477b)

3. C xS|Sx: oft Scyld Scēfing, “Scyld, son of Scef, often” (4a)

4. D S|Ssx: wīs wēlþungen, “wise, accomplished” (1927a)

5. E Ssx|S: glēomannes gyd, “musician’s song” (1160a)

The “Five-Type System” has been Sievers’s most influential legacy. It is the aspect of his theory that most nearly supports an understanding of the metrical system as accentual, stress-based, and thus homologous with later English meters. Yet the Five Types may not be the most important aspect of Sievers’s theory, or even the most reliable one. As Sievers also noted, the stress-based patterns he identified each unfold within a frame of four metrical positions. He termed these Glieder, or “members”; in the simplest realizations, each Glied is realized by a single syllable.⁵⁶ This is the basis from which Yakovlev builds his theory.

Prior to Yakovlev, metrists in the Sieversian tradition have always moved from the definition of four positions to definition of the stress-based rhythms or prosodic contours that form across them. Indeed, the great bulk of work within Sieversian metrics has been directed into cataloging and tabulating the various unique combinations of long and short, stressed, half-stressed, and unstressed syllables that occur in Old English verse – the Five Types and their forest of sub-types. Within this research program, the notion of the metrical Glied withdraws into the background; to the extent that it maintains a continuous presence, it is just a unit of verse possessed of a certain level of stress and which, in sequence with the other Glieder, forms the overall stress contour of the verse. Scansion aims to represent this contour. From here, it is just a short step to dispense with the metrical abstraction of Glieder altogether and treat the syllable as the definitive constituent. A. J. Bliss, who took this step, identified 130 unique contours among the 6,342 non-extended verses of Beowulf.⁵⁷ His alphanumeric notation is almost as intricate as the system of references to articles of the Summa theologica. (The first four verses of Beowulf are logged as types d3b, d3a, 1D1, and 1A1a, respectively.) Against this taxonomic impulse, Cable has always urged the principle of metrical simplicity. In important contributions that form key precursors to Yakovlev’s theory, Cable emphasizes that the meter’s “general principle” is its frame of four positions; he also demonstrates that the Five Types are just the contours that may occur, provided a small number of additional restrictions, within this four-position frame.⁵⁸ Yet even Cable turns from the definition of metrical positions to the definition of contours. By rejecting Sievers’s notion of metrical foot and challenging the notion of clashing stress, Cable has sought to bring the Five Types into closer agreement with intonational phonology. From one end of the Sieversian tradition to the other, researchers have interpreted their central task to be the correct description of stress-based prosodic contours. Yakovlev’s insight is that the stress contours may be epiphenomenal. Stress accent establishes the place of alliteration within the line, but it does not follow that the meter itself is stress-based, for alliteration may be only a regularized highlighting of stress peaks that are themselves the output of metrical organization located in a deeper linguistic layer.

To explain how linguistic material maps onto the basic four-position frame, Yakovlev posits that the meter distinguished between three types of morpho-syllabic constituent.⁵⁹ One class of morphemes contributes strong metrical positions, a second contributes either a strong or a weak metrical position, and the third, either a weak position or none. The third class consists of unstressed prefixes and ne: these morphemes may optionally go uncounted by the meter; when they are counted, they count as weak. The first and second classes are as one would expect from previous theories of Old English meter in the tradition of Sievers, except that Yakovlev dispenses with the distinction between primary, secondary, and tertiary stress. A morpheme with any of these three grades of stress forms a strong position; accordingly, the first class consists of “(the long syllables or resolved sequences of) roots, suffixes and stressed prefixes of open-class words, excluding finite lexical verbs.” The second class of morphemes consists of “inflections, unstressed prefixes, finite lexical verbs, and closed-class words.”⁶⁰

Ambivalence in the function of the second morpheme class (“either a strong or a weak metrical position”) is due to well-recognized phenomena of contextual promotion. When placed at the right edge of the verse, the root syllable of a finite lexical verb or closed-class word is promoted to contribute a strong position.⁶¹ Even the -i- and -od- of class 2 weak verbs (e.g., þancian, pret. 3 sg. þancode) make position independently in the coda.⁶² A different kind of metrical promotion evidently occurs verse-initially, in the absence of a word from the first class in the first half of the verse.⁶³ The promotability of morphemes – that is, the claim that an identical morpheme may contribute a weak position in one metrical context and a strong position in another – is simply carried over by Yakovlev from Sieversian metrics and should not be controversial. His innovation is instead to reduce the three (or four) stress grades into a binary distinction between strong and weak.

Nor is Yakovlev’s simplification of metrical stress wholly unprecedented. It receives support from Fulk, who urges that researchers have not distinguished sharply enough between metrical and phonological stress in Old English. Primary, secondary, and tertiary stress are necessary and justified in description of Old English phonology; at the metrical level, however, “it appears to be possible to simplify the description and reduce the number of levels of stress required, perhaps even to two – that is, stress and no stress.”⁶⁴ In Fulk’s assessment, “ictus at the tertiary level apparently amounts to syllable length” in combination with metrical context, and may therefore be described without reference to stress; meanwhile, “[t]he terms primary and secondary as applied to stress are convenient, but the distinction seems to have little to do with metrical realities,” for “secondary stress can be distinguished from primary on a purely positional basis: it is any full stress that immediately follows another full stress.”⁶⁵ Fulk’s analysis here is based in the principle of simplicity: the distinction between primary and secondary stress is derivable from general principles of Old English phonology; accordingly, it does not need to be repeated in a description of the Old English meter, for which a simple binary distinction will suffice.

Fulk does not pursue this line of thinking, but it supports his suggestion, quoted at the beginning of this chapter, that the role of stress in Old English meter has been overestimated. Indeed, some sort of theoretical adjustment quickly becomes necessary, for the simplification that Fulk and Yakovlev propose has the following unavoidable consequence: the number of metrical constituents designated as “stressed” increases massively. Consider, for example, the following two verses:

In a traditional Sieversian scansion, these verses are recognized as having two major stresses each: (9) is registered as Type D, S|Ssx, with suspension of resolution in –cyning; (10) is registered as a variant of Type A, Sx|Sx, with secondary stress in the dips. In the simplified system hinted at by Fulk and elaborated by Yakovlev, the scansion of (9) and (10) is, by contrast, SSSS in both cases. That improbable scansion – four consecutive “stresses” – is the direct consequence of treating primary, secondary, and tertiary stress as metrically undifferentiated: the tertiary stress of the suffix –ing, and the secondary stresses of the second elements in the compounds lēodcyning, glædman, and Hrōðgār are all counted as “stresses” in the simplified Fulk–Yakovlev system. Yet, the scansion SSSS is only unacceptable as long as we conceive of the Old English meter as stress-based, that is, a system keyed to contours of stress. The Sieversian scansions of (9) and (10) deliver reasonably accurate representations of the stress contours of these verses. The Yakovlevian scansion has a different representational target: it aims to represent metrical structure.

The shift in representational target entails a shift in the meaning of scansion symbols; hence my scare quotes around “stresses” in the previous paragraph. In Yakovlev’s system, the symbols “S” and “x” should be taken to designate strong and weak metrical constituents, respectively, not stressed and unstressed syllables. Here we return to Sievers’s concept of the Glied, his most significant contribution to Old English metrics. Verses consist of four Glieder, or metrical constituents, in any combination of strong and weak. To this, one must add that weak syllables adjacent to one another always combine to form a single Glied. There are accordingly eight possible permutations of strong and weak Glieder: SxSx, xSxS, xSSx, SSSx, SSxS, SxSS, xSSS, and SSSS. Strong positions are always formed by a single syllable or its resolved equivalent. Weak positions in the first half of the verse are expandable, but those in the second half of the verse (that is, the third or the fourth position) are formed by a single syllable. Where a dip in the third or fourth position is polysyllabic, it is usually reducible by the prefix license.⁶⁶ If the notation “(…)” is used for optional expansion of a dip, then the classificatory component of Yakovlev’s theory may be illustrated as follows.⁶⁷ Examples are from Beowulf. Where I give two examples, the second has a prefix uncounted by the meter:

1. I. Sx(…)Sx 11a “gomban gyldan”; 1322a “ne frīn þū æfter sǣlum”

2. II. x(…)SxS 477b “hīe wyrd forswēop”; 74a “Ðā ic wīde gefrægn”

3. III. x(…)SSx 13b “þone God sende”; 1223b “swā sǣ bebūgeð”

4. IV. SSSx 1927a “wīs wēlþungen”; 2930a “ābrēot brimwīsan”

5. V. SSxS 1160a “glēomannes gyd”; 5b “meodosetla oftēah”

6. VI. Sx(…)SS 17a “wuldres wealdend”; 217a “Gewāt þā ofer wǣgholm”

7. VII. x(…)SSS 4a “oft Scyld Scēfing”

8. VIII. SSSS 54a “lēof lēodcyning”

These permutations cut diagonally across the Sieversian Five Types: the members of Yakovlev’s first category are invariably Type A, but Type A verses with secondary stress belong to Yakovlev’s categories IV, VI, and VIII. The decisive factor is not the stress contour, but rather the quality of the linguistic material contributing each successive metrical position. The phenomena documented by accentual metrics are assimilated, transposed, and revalued – in a word, aufgehoben – within a non-accentual theory of meter.

Yakovlev lays out refinements to this system in his thesis, where he is careful to note that the accentual principle is not absent. The phenomena of alliterative patterning and contextual promotion show that “the accentual principle, with its attention to the general context of a phrase … is present in the background of the metrical system.”⁶⁸ This fact has important implications for later stages of the alliterative meter. Another set of diachronic implications may follow from Yakovlev’s interpretation of persistent asystematic patterns, of which five-position verses (Sievers Types A* and D*) are the most significant. These he describes as the “historical residue” of a prior metrical configuration: “a traditional metre,” he writes, “is hardly ever given opportunity to become completely cohesive. The average time span between major prosodic upheavals appears to be less than that required to eliminate any remains of the previous restructuring.”⁶⁹ This observation and Yakovlev’s accompanying analysis may prove to be of considerable interest for metrical paleontology, that is, reconstruction of pre-historical states of Old English meter. I limit myself, however, to two points at the close of this present chapter. One concerns historical evaluation of the accentual paradigm. The other concerns implications of Yakovlev’s analysis for later stages of the alliterative meter in English.

The Accentual Paradigm: Retrospect and Prospect

Students are right to be perplexed. The Old English meter is unlike anything that has followed it in this language; the profound differences from later English meters could only ever be grasped incompletely, and confusedly, within the accentual paradigm, which did its best to ignore them. The accentual paradigm originated as an effort to theorize meter as a direct, unmediated expression of language. Stress accent was what English could justly claim as its own, and this, as the inner genius of the language, was the truth expressed in its meter. Thus stated, the accentual paradigm is a recognizably nineteenth-century formulation, exhibiting the structures of thought characteristic of that period: the inner natures of things may be ontologically withdrawn, but the objects given to us in experience are nevertheless related to those inner natures as their expression.⁷⁰ In an effort to grasp the genius of English prosody, scholarship of the early nineteenth century rejected the normative force of Greco-Roman metrics and attempted to comprehend all English verse, as such, under a uniform principle of accentual rhythm. Researchers sought to bracket, as inessential, both the immanent vagaries of history and the spectral exteriority of norms. Against those purifying moves, one should acknowledge that metrical systems are inherently normative and that history is inherently digressive. Meter is not tidied-up speech and it cannot be deduced from linguistic phonology alone, for it has its own aesthetic norms and is shaped by its own historical course.

More specifically, the presence of stress accent in a language does not necessitate that its meter(s) be stress-based. Roman Jakobson’s justly famous formulation, to the effect that metrical prosody depends on linguistic prosody, does not allow us to deduce the former from the latter, for a system of versification only ever imposes its definitive equivalences upon a limited subset of the prosodic attributes available in the language. “If the system of versification is an unknown X and we are given nothing more than the prosodic features of the language,” Jakobson wrote, “we obtain an indeterminate equation, that is, the possibility that X could have more than one value”; though “the array of conceivable solutions” will always be limited, any actual system of versification represents only a single solution to an equation that might have been solved in another way, by fixing upon a different bundle of prosodic attributes.⁷¹

The case of classical Latin and its medieval reception may be instructive here. Though classical Latin probably had a stress accent, this did not prevent poets from writing in quantitative meters.⁷² When vocalic quantities were leveled in late antiquity, quantitative meters were sustained in the schools; simultaneously, new metrical systems were developed, keyed to the word-stresses that, though always present in the language, had played a secondary role in the classical quantitative meters. A prosodic feature subordinate within the quantitative meters was now established as the basis of a new metrical system. Indeed, some of the new stress-based meters had their origin in the “misreading” of quantitative forms, attending to the patterning of stresses rather than the patterning of syllabic quantity.⁷³ Following these medieval readers, it would be possible to make an accentual scansion of the Aeneid, marking the primary and, where relevant, secondary accents of each prosodic word in each line of Virgil’s poem:⁷⁴

This, it seems, is approximately what Bliss did for Beowulf: he produced an exacting record of a linguistic feature that is certainly present in the verse, and not without interest, but is epiphenomenal to the meter. In the Aeneid and other classical Latin verse, the organizing feature was syllabic length, matched to metrical positions designated by grammarians as feet; there was also some attention to word boundaries. In Beowulf and other Old English verse, the relevant features were syllabic length and morphological category membership, matched to metrical positions designated by Sievers as Glieder; there were also template-driven phenomena of metrical promotion, operative at key points in the line. The effects of stress contour may be perceived in various aspects of the Old English meter, and in Latin quantitative meters as well, but stress was only a subsidiary consideration in Latin meters of the classical period, and may not have been primary in Old English, either.

The accentual germ – whose presence in the Old English meter is already evident – gradually became the dominant principle. A morphological meter developed gradually into an accentual one, but traces of the morphological precursor persisted within the accentual successor, as embedded reflexes. The result is that some features of even the Middle English alliterative meter remain beyond the grasp of the accentual paradigm. Yakovlev’s historical perspective suggests solutions to several long-standing problems – or, better, it shows that the problems are problems in re, in the Middle English meter itself, and not merely in the modern understanding of it. The meter was negotiating its morphological inheritance. That inheritance will be investigated in the next three chapters of this book, but several points may be made at this juncture, as a caution against overextensions of English accentual-syllabic prosody. Recent work on the Middle English b-verse can seem to represent a historical vindication of the accentual paradigm: by defining meter in terms of stress contours, the accentual paradigm always implied that unstressed syllables ought somehow to be regulated, and Cable and Duggan showed that this was indeed the case in the second half of the Middle English line. That triumph might lead one to think that other principles of English accentual-syllabic metrics – the principle of a fixed stress count, for example, or the monosyllable stress rule – should be applicable to fourteenth-century alliterative verse. Yet the situation is considerably more complex.

Consider, again, clashing stress. It is a rare and marked realization of accentual-syllabic meters, and runs counter to ordinary rhythms of speech articulation, but it occurs frequently in Middle English alliterative verse. It is, moreover, one of the normal realizations of the b-verse:

Why did an accentual meter make such persistent use of this difficult contour? Metricists have been right to flag the problem. For Yakovlev, however, the answer is simple: clashing stress is present within the Middle English meter as a reflex of its historical precursor.⁷⁶ The morphological meter specified only arrangements of strong and weak elements, allowing the contour to take shape within those constraints. Strong positions frequently adjoined one another; when they did, the result was often stress clash. As the accentual principle became stronger in the early Middle English period, patterns with stress clash would come to appear increasingly marked. They nevertheless remained part of the rhythmical repertoire.

Ambiguities in stress count may be explained along similar lines.⁷⁷ The morphological meter permitted a minimum of two and maximum of four strong constituents per verse. Heavy verses – that is, those with three or four strong constituents – would usually have only two stress peaks, as is manifestly the case in (9) and (10). Nevertheless, it seems likely that a “strong constituent” came to be interpretable as a stress during the transitional period, thereby establishing three-lift verses as an acceptable realization of the meter in Middle English. This speculation is at odds with current understandings of Old English heavy verses and their fate, but the matter merits reassessment. It is well established that the frequencies of Sievers Types D and E are sharply reduced in late Old English verse; since Oakden, progressive disappearance of Types D and E from the metrical repertoire has been tied to the decline in compounding.⁷⁸ Yet this line of reasoning may give too much credit to changes in poetic lexicon, and too little to the metrical system. Against the received view, one might suppose that Sievers Types D and E were deselected because they afforded no opportunity for a polysyllabic dip, which had become a requirement of the meter in the Middle English period. If the long dip requirement eliminated Sievers Types D and E from the repertoire, heavy verses (of which Types D and E are the most prominent representatives in the Sieversian notation) might be expected to survive in one form or another, for they were a central component of the metrical system. The relaxation of the four-position rule was decisive in this regard. The alliterative meter became more accommodating to unstressed syllables, which were in any case proliferating under the pressure of analytic syntax. The new infusion of unstressed syllables spaced out strong positions, while the shift from morphological to accentual organization meant that each strong position was increasingly supplied by the lexical stress of an independent word. The result was three-lift verses. Three-lift verses in Middle English are probably best understood as a reflex of an earlier configuration of the metrical system, one in which the meter did not count stresses, and permitted as many as four strong constituents per verse.

In the transition from a morphological meter to an accentual one, the distinction between grammatical word classes remained largely intact: the Middle English meter derived metrical stresses, or lifts, from the same categories of words that formed strong positions in the Old English meter. What changed in the later period was just that the metrical system became keyed to the lexical stresses of these words. Polysyllabic nouns, adjectives, and participles (words that may have contributed more than one strong position in Old English) now typically contributed only a single metrical stress, on the syllable bearing primary lexical stress. This transition to an accentual system was a step towards the accentual-syllabic prosody of later English verse, but only a single step in that direction. The next steps will be traced in the next three chapters of this book, and their conclusion was the breakdown of the alliterative meter as an independent mode of versification in English.

These are difficult problems, and my presentation of them is compressed. The first task, before offering a fuller reconstruction of changes in metrical system, is undoubtedly to establish that Old English poetry was indeed the historical progenitor of Middle English alliterative poetry, and that these two corpora belong to a continuous tradition of verse practice. This is where the next chapter begins.