Did Old English Verse Have a “Morphological” Metre?

Abstract

The revision of the four-position theory of Old English metre by Yakovlev (2008) has had a considerable impact, both for its simplification of Sievers’ (1893) metrical principles, and for its supposed shift to a “morphological” rather than an “accentual” metrical type. I contextualize Yakovlev’s important contribution to metrical theory, highlighting that his main innovations are to eliminate the principle that every verse should have two lifts, and to collapse the metrical notions of “lift” and of “half-lift” into a single type of unit, the “strong position”. These major and thoughtful innovations are unconnected to the supposed “morphological” aspect of Yakovlev’s system, which consists of arbitrary and unmotivated stipulations of certain classes of syllables as inherently strong or weak. Calling the metre “morphological” on this basis is at best misleading, placing too much emphasis on a marginal component of the system. Moreover, the definitions of strong and weak positions are better explained with reference to linguistic stress: elements bearing some degree of stress (primary or secondary) are strong, while those bearing no stress are weak. Such a reframing leads to a slight revision of Yakovlev’s theory to incorporate the “rule of the coda” (Fulk, 1992). The result is a version of Yakovlev’s theory which is both theoretically simpler and more descriptively adequate, but in which the label “morphological” lacks even the limited and inapt validity of the original version. This reintroduction of stress into the system does not, however, make the theory “accentual”, and it remains better characterized as an “alliterative-syllabic” view of the metre (Cable, 1991).

Introduction

In Goering (2023: 248, n. 15), I briefly commented on the inappropriateness of the label “morphological” for Yakovlev’s theory metre, and that I was preparing an article to discuss the issue at greater length. This is that article. I would like to thank the anonymous reviewer for Neophilologus for saving me from a number of errors and obscurities of presentation.

In an important doctoral dissertation, Yakovlev (2008) proposes a revision of the mainstream theory of Old English metre. This new system does not reinvent the wheel, and builds on a tradition of metrical research begun by Sievers (1885a,b, 1887, 1893), and refined especially by Cable (1974, 1991). In Cable’s revision, the metrical system used in Old English poetry has two really key basic principles: each half-line of poetry should have exactly four metrical positions (often a single syllable, but with rules in place to allow more syllables per position under various circumstances), and two of these positions should have a special metrical prominence (usually corresponding to linguistic stress) that makes them stand out as lifts. As explained in more detail below, Yakovlev’s contribution is nothing more or less than the elimination of the second of these principles: he does away with the notion of the two-lift requirement, leaving position-counting alone as the basis of the metre. The result is an elegant simplification of the system, though one that comes at the cost of explaining a few quirks of scansion slightly less well.

The nature of this contribution has, however, been presented rather differently – and, I argue here, inaccurately – in metrical scholarship since 2008, by both admirers and critics of Yakovlev’s theory of scansion. The problem starts already with Yakovlev himself, who describes his system as a “morphological metre” (2008: esp. 81–82). He offers this as an alternative to the common description of Old English (and cognate Germanic) metre as “accentual”, in keeping with his rejection of the two-lift (two-stress, two-accent) basis of the system.

This characterization is endorsed in two recent broad surveys of English alliterative verse. Weiskott (2016: 24–25) is the more cautious, describing the term morphological as “illuminating”, but rightly recognizing that Yakovlev has not, in fact, entirely done away with “accent” (stress) as a relevant feature:² it remains “a meter with at least three recognizable principles of organization: morphological, quantitative, and accentual” (2016: 25).³ Cornelius (2017: 57–62), by contrast, leans more heavily into the supposedly non-accentual nature of the theory, and calls Yakovlev’s proposal and terminology a “paradigm shift”, finding in it a (welcome) revolution in how we look at Old English verse.⁴ In a critique of the theory, Neidorf & Pascual (2020: 250) argue that “Yakovlev’s claim that the meter was fundamentally morphological … appears needlessly complex and incoherent”. Like Cornelius, they are heavily invested in Yakovlev’s label, to the extent that they present a critique of the supposed morphological aspect, and the role of stress in the metre (Neidorf & Pascual, 2020: 249–250), as part of a critique of the theory a whole.

My main purpose in writing this piece is to try and entirely excise the inaccurate and confusing term “morphological” from discussions of Yakovlev’s metrical analysis. His system is not a “morphological metre” in any useful sense. The inadequacy of this term, however, does not have any bearing on the merits or faults of the core of Yakovlev’s proposal: this label can be entirely dispensed with, and everything that is significant about his contribution remains untouched – or even strengthened.

In what follows, I first give a highly condensed outline of the metrical theories of Sievers, Cable, and Yakovlev, focusing on the ways that successive waves of theorizing have built on what came before. I then consider what a “morphological metre” might potentially mean, what Yakovlev seems to mean by the term, and why this is not a useful or helpful descriptor. In the subsequent section, I turn to the main alternative to Yakovlev’s morphological classifications, namely linguistic stress, which straightforwardly occupies the role that Yakovlev tries to claim for morphology. Adopting stress rather than morphology as a principle in defining metrical positions requires a rather minor adjustment to Yakovlev’s system, namely the incorporation of the long-recognized rule of the coda, a matter reviewed in the penultimate section. The result is a version of Yakovlev’s theory that is more robust and descriptively adequate – though, as I note in the conclusion, many questions remain about how to explain Old English metre, and how to judge competing theoretical descriptions.

A Brief History of Old English Metrical Theory

Sievers

The foundational figure in Old English metrics remains Eduard Sievers, and Yakovlev’s own theory is deeply rooted in this Sieversian tradition (Goering, 2020).⁵ Sievers approached Old English as one component in a wider spectrum of metres in various Germanic languages, and in a famous synthetic monograph, he treated Old Norse, Old English, Old Saxon, and Old High German together using a common system inflected and altered in the various traditions (1893). Though he gave due consideration to features like alliteration, his special contribution concerned the structure of verses (half-lines). Half-lines do not follow a single obvious pattern such as strict syllable-counting, regularly alternating stresses, or quantitative feet,⁶ but Sievers realized that they are also not entirely free in their composition. His work to understand the kinds of regulations at work consisted of essentially two broad aspects, which might be reasonably termed linguistic analysis and metrical analysis.

Linguistically, Sievers’ analyses have not always remained entirely uncontested, but all mainstream metrical work since has accepted, on some level, Sieversian principles such as resolution (and its conditioned suspension),⁷ or the occurrence of archaic linguistic forms retained for metrical convenience or poetic effect (especially metrical preservation of forms without contraction or epenthesis/“parasiting”). The current standard treatment of these features in Old English poetry is Fulk (1992), and I will mostly set them aside for the remainder of this discussion. Except in a few particulars, such as the details of Kaluza’s law (Goering, 2021a, 2023: ch. 5, both with literature), they now involve few serious controversies. They are part of the metrical-linguistic landscape which any good metrical theory needs to describe, not the special preserve of any specific metrical framework.

Of more immediate relevance is Sievers’ metrical theory: that is, his elucidation of the metrical rules that determine what verse-patterns are acceptable, which ones are prohibited, and which ones are only permitted under special licences or circumstances. The original version of Sievers’ theory was rather cumbersome and overburdened, involving three main principles:

• 1. Each verse should have four metrical positions.

• 2. Each verse should have two lifts.

• 3. Each verse should have two feet.

The definition of metrical positions is an essential aspect of the system, with Sievers identifying three distinct types of position:

Lifts The strongest type. A lift prototypically consists of a single syllable (or resolved equivalent) carrying primary linguistic stress,⁸ and participates in the alliterative scheme of the line. (German: hebungen)

Half-lifts Of middle strength. A half-lift prototypically consists of a single syllable (or resolved equivalent) carrying secondary linguistic stress, and does not count as part of the alliterative scheme. (German: nebenhebungen)

Dips The weakest type of position. A dip consists of one or more syllables of weaker or no stress linguistically. (Also called drops; German: senkungen)

A crucial point here is that where lifts and half-lifts usually have a single syllable (or at most two, by resolution), dips frequently have several weak syllables in a row, all counting as a single metrical position. One consequence of this is that two dips cannot occur next to one another, since all the weak syllables in a row would automatically be parsed as a single dip. This can be indicated more schematically, representing lifts by S, half-lifts by s, and weak syllables and dips by w.⁹ Using this notation, a verse of the pattern SwwS (lift–dip–dip–lift) would be impossible, since a linguistic contour of this shape would be scanned as metrically SwS.

Using these definitions of metrical positions and lifts, along with a further (and rather arbitrary) grouping of positions into feet, Sievers (1893: 31) could present his famous five types of Germanic verse, with each type being given a letter label based on its frequency of occurrence in Old English:

• A Sw|Sw

• B wS|wS

• C wS|Sw

• D S|Ssw or S|Sws

• E Ssw|S or Sws|S

These five types were expanded to a couple of dozen subtypes, with labels such as A2ab, C3, and D*4 representing various details such as the use of a half-lift in place of a dip, certain details of resolution, or the presence of other metrical permutations. This system of types and subtypes proved very influential, and a good deal of later metrical research – especially the well-known monograph by Bliss (1962) – has seemed to take the goal of metrical research to be the elaboration of types and labels.¹⁰ For many decades after Sievers, much attention was given to what types were found in verse, but little to why some types should be allowed and others not.

Cable

A key advance in metrical theory was to come from Cable (1974). He dispensed with Sievers’ rather confusing system of feet.¹¹ This left just two principles at work in the metrical system:

• 1. Each verse should have four metrical positions.

• 2. Each verse should have two lifts.

From these two principles, along with Sievers’ definitions of lifts, half-lifts, and dips, there are six possible logical combinations of two lifts and two non-lift positions. Illustrating the options using S for the two lifts and n for the non-lift positions (dips or half-lifts), the six options are:¹²

A SnSn

B nSnS

C nSSn

D SSnn

E SnnS

X nnSS

In types A–C, both the non-lift (n) positions are by default dips, though replacement of dips with half-lifts is common in type A. In types D and E, one of the non-lift positions must be a half-lift, for the simple reason that two adjacent dips were no more allowed under Cable than they had been under Sievers – SSww would count as SSw, and SwwS as SwS. Since those patterns would be a verse-position short, only the combinations SSsw (Da), SSws (Db), and SswS (E) are possible.¹³ A half-lift cannot be introduced in the sixth option, however, since the linguistic structure of older Germanic languages does not allow for a half-lift to occur before a full lift: a hypothetical pattern such as ˣswSS or ˣwsSS cannot occur for linguistic reasons. Since wwSS would count as three-position ˣwSS, only the first five types are valid under Cable’s system. In other words, Cable spelled out the underlying logic that explains why Sievers’ system had exactly five basic types: this is not an arbitrary number, but an inevitable result of the principles of four positions and two lifts.

One important theoretical consequence of Cable’s revision is that even as he gave more rigour and justification to the five types, the types themselves became almost unimportant except as a descriptive shorthand.¹⁴ Sievers had seemed to consider the five types to have some kind of fundamental reality in and of themselves. He even accepted an analysis of hypermetric verses as blends of two types (Sievers 1893: 139), suggesting he thought Old English poets used the types as compositional building blocks. In the conclusion of his 1974 monograph, Cable – perhaps merely as an expositional device – seemed to present his five types in much the same way, but some years later he was explicit that the types must be mere epiphenomena (1991: 37–40). It is the basic principles that generate them that really matter, not the precise number or shape of the types as such.

Yakovlev

Yakovlev’s contribution to this theoretical tradition was to eliminate the two-lift requirement, while retaining the core apparatus of the four-position approach. Under Sievers’ system, lifts and half-lifts had been defined in the same way in terms of how linguistic material maps onto metrical positions. Both normally consist of either a single (partly or fully) stressed syllable, or of two syllables that are resolved together. Cable (1974: ch. 6) had indeed already noted that there can sometimes be a great deal of ambiguity in distinguishing lifts and half-lifts, especially in verses of the following type:

(1) scyld wēl ge·bearg

‘shield protected well’ (Beowulf 2570b)

Is this type E, SswS, or type Db, SSws (Cable, 1974: 77)? Both scansions have been proposed in the scholarly literature, and Cable rightly observed that, as long as alliteration is not at stake, the distinction hardly matters for any practical purpose. Yakovlev took this reasoning one step further, and he made no distinction between lifts and half-lifts at all. Instead, he identified just one kind of strong position, contrasting with dips/weak positions.

Doing away with the two-lift principle left Yakovlev with a single fundamental metrical rule:¹⁵

1. Each verse should have four metrical positions.

Beowulf 2570b would then be notated as SSwS. It is not that the verse has three lifts, but that the concept of “lift” is irrelevant. There are four metrical positions, three of which happen to be strong positions. There is no need, Yakovlev argues, to elevate two (and only two) of those strong positions as lifts while leaving the third as a half-lift – in metrical terms, all strong positions are equivalent (though they will usually have varying prominence linguistically).¹⁶

As Yakovlev (2008: 74–75) noted, this principle, along with his revised definitions of strong and weak positions, leads to eight types rather than five:¹⁷

• 1. SwSw

• 2. wSwS

• 3. wSSw

• 4. SSSw

• 5. wSSS

• 6. SwSS

• 7. SSwS

• 8. SSSS

All other possible arrangements would involve adjacent weak positions, which are still as impossible for Yakovlev as they had been for Sievers and Cable. But, just as for Cable, these types are epiphenomenal, and are at best a notational convenience for metrical scholars rather than being a description of the metre per se. It is also worth noting that while Yakovlev’s principles generate slightly more basic types, he does not work with subtypes at all. This makes sense, since the majority of subtype classifications developed by Sievers had been indications of the use of half-lifts and (non-)resolution, neither of which are indicated explicitly in Yakovlev’s notation or typology.

Presented this way, Yakovlev’s metrical theory does not really seem to represent a “paradigm shift” so much as an insightful and elegant refinement. Whether it is in all ways an improvement remains to be seen, but on a theoretical level, this system has far more similarities with than differences from its predecessors. This high level of continuity naturally raises the question of whether Cable’s system was really meaningfully “accentual”, and whether Yakovlev’s revision has really changed it to a new “morphological” type.

What Would a “Morphological” Metre Be?

The aptness or otherwise of the term “morphological metre” of course depends on what such a label might convey. Yakovlev’s desire to avoid the term “accentual metre” is obvious enough: his main move is to get rid of the two-lift principle, which has been widely understood as resting on a system that essentially required two main linguistic accents per verse – even if many lifts did not, in the event, correspond strictly to linguistic primary stresses. In other words, “accentual metre” should be taken, in this context, as meaning “a metre which counts two accents in each verse”.

Such a definition of accentual metre might perhaps seem to imply that a “morphological metre” would in some way count morphological entities.¹⁸ This is clearly not what Yakovlev means by the term, however. Old English verses can, very obviously, vary greatly in terms of the number of morphemes they contain, and no one – certainly not a metricist as sensitive as Yakovlev – could take it for a “morpheme-counting metre” in any sense.

Instead, Yakovlev’s use of “morphological” refers to the mapping of linguistic material onto metrical positions. His judgements of what specific syllables can fill strong and weak positions correspond in practice almost exactly to those universally accepted in Sieversian metrics, but his theoretical definitions of these positions is very different, and worth quoting in full. For strong positions, Yakovlev (2008: 75, no. 17) identifies the following syllables:

Strong metrical positions are formed by (the long syllables or resolved sequences of) roots, suffixes¹⁹ and stressed prefixes of open-class words, excluding finite lexical verbs; strong metrical positions are also formed by (the initial – except for an unstressed prefix – long syllable or resolved sequences of) any other word displaced from its normal syntactic position and/or standing verse-finally.

And for weak positions, he allows the following material (2008: 75, no. 18):

Weak metrical positions are formed by (the syllables of) inflections, unstressed prefixes, finite lexical verbs, and closed-class words.

In other words, a syllable’s status as strong or weak is predetermined on a morphological basis (except when it isn’t, as in the case of syntactic displacement). This is apparently what is meant to make the system “morphological”.

There are at least three major problems with resting the fundamental characterization of Old English metre as “morphological” on these definitions. First of all is that Yakovlev’s redefinitions are entirely unrelated to his theory in other respects. He could have defined metrical positions purely in terms of stress, and no other aspect of his larger theory would be altered. Yakovlev could have introduced his key innovation – that there is no rhythmic distinction between full lifts and half-lifts – without touching on the notion of morphology. Or he could have made a much more modest revision of Cable’s system which retained lifts and half-lifts, but which defined these “morphologically”: certain syllables would be inherently strong, and only eligible to be lifts or half-lifts, while others would be generally weak, and only eligible to stand in dips (unless syntactically displaced, etc.). Such a morphological definition of syllable strength should not seem much more strange within Cable’s system than it does within Yakovlev’s. Whether candidacy for stronger metrical positions is determined by stress or morphology is an issue that is, strictly speaking, orthogonal to whether there are two kinds of strong position or just one.

Second, it is hard to see how this really makes the metre overall genuinely “morphological”: it remains a position-counting metre. If positions are instead defined in terms of stress, would this make the metre suddenly become “accentual”? Hardly: it is still metrical positions that are counted, and under Yakovlev’s own formulation, two to four of these could be strong, with zero to two being weak. How syllables are mapped to metrical positions is important, but whether the metre is “accentual” or “morphological” or (as I would suggest) neither does not depend on this point. The best way to describe the four-position approach’s view of Old English metre comes from Cable (1991: 26): “Old English meter is alliterative-syllabic, containing four positions”. Neither accent nor “morphology” lie at the heart of the system, and using either term places undue emphasis on that aspect.

The third criticism of Yakovlev’s idea of a “morphological” metre is that it is largely redundant with, and in some ways inferior to, a much simpler definition of strong positions in terms of stress. This point is more involved, and deserves its own section.

Stress

Yakovlev’s definitions of what syllables count metrically as inherently strong and which ones are usually weak, given in the previous section, are long and arbitrary. There is, however, one linguistic feature in Old English which does, with a few exceptions, group the elements Yakovlev wants to see as strong together: stress. Yakovlev even has to invoke stress directly (“stressed prefixes of open-class words”), and while it is hard to see why syntactic displacement should have any effect on an element’s morphological status, it is not surprising that this should affect the phrasal stress of some elements (Neidorf & Pascual, 2020: 249–250).

Within words, linguistic stress also corresponds well (though not quite perfectly) to Yakovlev’s listing of strong elements. The roots of lexical words (excluding finite verbs in some positions) are generally stressed in older Germanic languages, and many nominal prefixes are shown by the alliteration to also carry stress. For these elements, linguistic stress predicts Yakovlev’s metrical strength exactly.

Old English also allows secondary stresses within words: as traditionally understood, this falls on heavy medial syllables,²⁰ and the lexical roots of later elements within compound words (Hutton, 1998; Goering, 2023: ch. 4–5).²¹ It would be easy to use this linguistic basis for stress to propose much simpler rules for mapping syllables into strong and weak positions:

• 1. Strong metrical positions are filled by syllables (or resolved equivalents) bearing primary or secondary stress.

• 2. Weak metrical positions are filled by syllables bearing weak or zero stress.

There are some mismatches between the elements counted as strong or weak these revised rules and the judgements made by Yakovlev’s system. This goes both ways: some elements that Yakovlev’s rules would count as weak bear secondary stress linguistically, and some that Yakovlev considers strong should be linguistically unstressed. In all cases of discrepancy, the stress-based rules provide a better fit with the metrical system of Beowulf than do Yakovlev’s classifications.²²

For an example of where a stress-based definition of metrical strength works better, take the adjectival suffix -ig. By Yakovlev’s rules, this suffix ought to simply count as metrically strong, without exception, whenever it occurs. This is problematic in certain verses, such as the following:

(2) ǣnig ofer eorþan

‘any on earth’ (Beowulf 802a)

(3) mōdig on ge·monge

‘high-spirited in company’ (Beowulf 1643a)

If the -ig were ‘strong’, metrically speaking, then the scansion would seem to be SSwwSw. This is not a valid verse-type,²³ while the alternative scansion SwwwSw is a robust and well-attested pattern (with the normal four positions that Yakovlev’s system takes as normative).

On the other hand, there are contexts in which -ig should count as metrically strong:

(4) sārigne sang

‘sorrowful song’ (Beowulf 2447a)

(5) drīorigne fand

‘found stained with blood’ (Beowulf 2789b)

Here the scansion must be SSwS, since SwwS would be reduce to SwS: two adjacent weak syllables form a single dip, which would leave the verse having just three metrical positions.

Under Yakovlev’s rules, this kind of variation is inexplicable. It must be regarded as, at best, a licence that poets can exploit for their convenience (2008: 78). But early poets are not, in fact, able to treat -ig however they wish. Final -ig never needs to count as metrically strong, while (at least in Beowulf and similar poems) -ig- always counts as strong when followed by a consonant-initial suffix, such as -ne (Fulk, 1992: 214–215). From a stress-based viewpoint, this is straightforward: a heavy medial syllable like those of sā́rìgne or drī́orìgne automatically takes secondary stress, and so must count as metrically strong, while a word-final suffix ending in a single consonant always counts as metrically weak.²⁴ Yakovlev cannot capture the systematic nature of this alternation in older poetry without (explicitly or covertly) introducing stress as a factor.

Further problems arise with light medial syllables. Yakovlev (2008: 76) considers elements such as the vocalic stem of class II weak verbs to be inherently strong, metrically, though linguistically the expectation is that they are fully and invariably unstressed (Fulk, 1992: 203–204, Russom, 2001: 45–46, Stausland Johnsen, 2015). This creates problems with verses such as the following:

(6) þrēatedon þearle

‘attacked fiercely’ (Beowulf 560a)

(7) geōmrode giddum

‘lamented with songs’ (Beowulf 1118a)

(8) sweðrian syððan

‘subside afterwards’ (Beowulf 2702a)

All of these verses would have a five-position scansion, SSwSw, by Yakovlev’s rules. By the expectations of linguistic stress, the much more welcome scansion SwwSw falls out automatically. There is in Beowulf only one verse in which it might seem preferable to scan a class II weak verb with stress on the stem:

(9) egsode eorl

‘he terrified a warrior (warriors? the Heruli?)’ (Beowulf 6a)

But this verse is poor evidence, being not only entirely isolated within the poem, but also very likely corrupt and to be emended (Drout & Goering, 2020).

Yakovlev does not address these specific verses, though he does try to explain the suffix scipe, which is likely also unstressed, as undergoing resolution when needed (2008: 76, n. 49). That is, verses like Beowulf 560a (example 6) might perhaps be taken as SSʷSw (using the superscript w to highlight – for the sake of exposition – that the position is a resolved one). But this would require very many violations of Kaluza’s law, which Yakovlev is quite rightly at pains to incorporate into his system. Put another way, the evidence of Kaluza’s law is a further indicator that the stem of class II weak verbs is not stressed.²⁵

In every point where Yakovlev’s rules and a stress-based view of the metrical system conflict, the stress-based approach makes the better predictions about which elements are to be scanned as strong and which as weak.²⁶ The obvious conclusion is that Yakovlev’s long and arbitrary lists of “morphological” categories (plus syntactic displacement) should be replaced by the simple stress-based rules given at the start of this section. As noted in the previous section, such definitions do not suddenly make Yakovlev’s metre “accentual”; it remains position-counting, with positions reducing in many cases to syllables, however position strength is defined.

The Rule of the Coda

The claim I made in the previous section, that the stress-based model always predicts better which metrical positions are strong and which are weak, might at first glance seem to be contradicted by verses such as:²⁷

(10) swā rīxode

‘thus ruled’ (Beowulf 144a)

This alliterates on r-, and swā is universally acknowledged to map to a weak position. Yakovlev (2008: 76) cites this as an example of suspended resolution, implying – as one would expect from his general rules – a scansion wSSw. Understanding the second syllable of rīxode as uniformly unstressed would give a scansion of wSww, which under Yakovlev’s rules as stated is unacceptable: two adjacent weak positions ought to collapse into one, making this a three-position verse.

The problem here is not with the stress-based model, but with a minor deficiency concerning the metrical mapping in Yakovlev’s theory. The behaviour of class II weak verbs (and all other words of similar shape) has been essentially explained by Cable (1991: 19) and especially Fulk (1992: ch. 7, esp. 201), through what Fulk termed “rule of the coda”. Fulk’s explanation is that when a verse ends in two unstressed syllables, such as those in rīxode (Sww), the first of these will receive metrical ictus (this is not necessarily the same as linguistic stress). In other words, he proposes a rule that promotes a linguistic contour such as wSww to the metrical contour wSSw. This rule was framed as it was largely due to Fulk’s commitment to the principle that every verse should have two lifts (and a lift is, by definition, a metrical ictus).

Within Yakovlev’s framework, which rejects the two-lift principle, the rule of the coda may be stated much more efficiently, to the benefit both of the rule itself and of Yakovlev’s system overall. The rule that adjacent weak syllables should collapse into a single weak position becomes:

Non-verse-final weak syllables count as a single metrical position.²⁸

A rule like this is in keeping with the general principle of closure, which holds that the mapping of linguistic elements to metrical units will, generally, become stricter and more precise towards the end of a verse or line (Hayes, 1983: 373, Russom, 2017: 17).

Adopting the rule of the coda makes two further types possible within Yakovlev’s system:

1. wSww

2. SSww

Both types occur. The former has already been illustrated in swā rīxode, while the latter is well-attested as part of the traditional type D:²⁹

(11) stīg wīsode

‘the path guided’ (Beowulf 320b)

The addition of two more types or configurations of strongs and weaks to the system is not a problem: as noted earlier, the exact number of types is entirely irrelevant, since they are purely epiphenomenal. The only thing that matters is the principles generating the types, and the adequacy with which the system describes the data. Incorporating the rule of the coda into Yakovlev’s framework is a very mild complication (modifying the principle of dip-formation in a way very much in keeping with universal metrical trends), and it allows the mapping of linguistic and metrical elements to be considerably improved, increasing the descriptive adequacy of the system. There is no need to invoke a long and arbitrary list of morphemes that count as strong or weak, and the usual rules of linguistic stress, interacting with a very slightly modified version of Yakovlev’s existing principles, are sufficient.

Conclusion

Debate about the metrical fundamentals of Old English – and, it should not be forgotten, related Germanic – metre will no doubt continue. It may seem is if I am weighing in on this debate on the side of Yakovlev, and am a “proponent” of his system. Insofar as I feel that Yakovlev’s theory has been incorrectly characterized (beginning with Yakovlev himself), and that some responses, criticisms and acclamations alike, have followed from these misapprehensions, I suppose that I am offering a defence of sorts. I am not, however, offering an endorsement. There are still some problems with Yakovlev’s system, as there are indeed with all versions of the four-position theory currently proposed.

Some of these problems were raised by Neidorf & Pascual (2020), though their criticisms are not in all ways fully accurate, and many apply even to their own favoured version of the four-position system (which is essentially Cable’s two-lift, four-position approach, distinguishing full lifts from half-lifts). The most obvious of these problems concerns Yakovlev’s hypothetical wSSS configuration.

Discussion about this type has been muddied first of all by the empirical question of whether such a type does or does not occur. Yakovlev (2008: 75) assumes that wSSS is found, and provides what he claims is an example:

(12) syððan Hygelāc læg

‘after Hygelac lay (dead)’ (Beowulf 2201b)

Neidorf & Pascual (2020: 248) are, however, surely right that the second part of the name Hygelāc should here, as often elsewhere, be taken as fully unstressed (see further Pascual, 2020): this particular example scans as wSwS, a common pattern (Sievers’ type B). There are indeed very few verses ending in three elements with uncontroversial stress, such as (to use Neidorf & Pascual’s example):

(13) ˣand heal-ðegn wæs

‘and was a hall-thane’

I can find only one example from early verse that seems to show the wSSS pattern:

(14) ðā-ði geolu god-ueb

‘those who (consume) the yellow divine cloth’ (Leiden Riddle 10a)³⁰

This should have the linguistic contour wwSʷSs, which is for Yakovlev metrically wSSS – but this verse is conspicuous in its lack of parallels, and Hutcheson (1995: 33, n. 120) terms it “unmetrical”. In other words, as far as the empirical question concerning the existence of the wSSS type goes, Neidorf & Pascual clearly have the right of it, and this is indeed a point where Yakovlev’s predictions fail to match reality.

The non-existence of wSsS and wSSs verses is, however, also something that two-lift theories need to account for. Cable (1991: 148–151) was aware of this problem. On the normal, basic principles, there is no reason why poets should not have composed verses of the shape SwsS (“E2”), wSsS (B with a half-lift instead of a second dip, such as the hypothetical example 13), or wSSs (C with a half-lift instead of a second dip; this is the type Leiden 10a represents). All these verses have two lifts, and involve what would seem to be routine realization of a non-lift position by a half-lift. Their absence is as problematic for Cable as it is for Yakovlev: on this point, the criticism of Neidorf & Pascual is a double-edged sword, applying equally to both classic and revised four-position approaches.³¹

This is not to say that such criticisms need be damning for either approach. Positional theories, whether Cable’s or Yakovlev’s, might invoke further principles beyond their core mechanisms to explain some of the non-occurring verse configurations. Cable (1991: 148–151), for instance, looks to stress-clash avoidance, proposing that a rising contour of wsS was prohibited: this explains the absence of the “E2” type ˣSwsS (though the wSsS and wSSs gaps are harder to explain so elegantly; Goering, 2023: 249–250, n. 19). This solution can also work for Yakovlev: as long as this is understood as an avoidance of a non-euphonic linguistic sequence rather than a restriction on the combination of metrical units, the absence of “E2” verses can be accounted for without needing to invoke the half-lift as a formal metrical unit. Just as potential criticisms of Yakovlev’s theory may have a wider applicability to four-positional approaches more generally, so too can at least some of the mechanisms designed to address such problems serve Yakovlev equally well.

It should also be noted that despite its current popularity, the four-position approach is not the only Sieversian theory with a real claim to theoretical attention. I personally am somewhat inclined to prefer the word-foot theory (see now Russom, 2022, with full references to his substantial range of publications on the topic) over the four-position framework.³² The word-foot model is theoretically more complex, but accounts better (in my view) both for some of the finer details of Old English metre – the absence of ˣSwsS, ˣwSsS, and ˣwSSs verses follows readily from its general principles, for instance – and can be applied to the full sweep of Germanic metres. Positional approaches tend to face difficulties when confronted with a poem such as the Norse Atlakviða,³³ and struggle to adequately regulate the ljóðaháttr metre, while the word-foot approach can be adapted to all these forms (see Goering, 2016b: ch. 4 on ljóðaháttr full-verses). More work remains to be done on many of these areas, but as it currently stands the word-foot theory appears to be a promising framework for explaining Germanic alliterative verse in general.

But I did not write this piece to specifically defend or reject any metrical approach. Rather, my main goal is to argue for keeping the terms of theoretical comparison fair. If Yakovlev’s approach is to be criticized or praised, this should be for real features of the system, not an excessive fixation on the ill-chosen and effectively irrelevant label “morphological”. Furthermore, if it is to be rejected, every attempt should be made to first see if the apparent flaws can be addressed. In particular, the lack of attention to the rule of the coda is not in itself grounds for setting Yakovlev’s entire system aside, since this rule can be so easily incorporated into the larger framework.

All in all, Yakovlev’s contribution to metrical theory is considerable. His rejection of the two-lift principle, and the concomitant collapsing of lifts and half-lifts into strong metrical positions, is in many respects an insightful and satisfying refinement and simplification of the four-position approach, and at the least highlights how much traditional lifts and half-lifts have in common. His approach is not flawless, but this is only to be expected, given the complexity of dealing with Germanic metrics in general. If his contribution is to be criticized, his achievements should still be acknowledged; if his system is praised, this should be done with clear eyes and an appreciation for its limitations.