On a Crux in Beowulf: The Alliteration of Finite Verbs and the Scribal Understanding of Metre

Abstract

Because of its faulty alliteration and irregular metrical configuration, the manuscript reading felasinniġne secg, corresponding to l. 1379a of Beowulf, has traditionally been emended to sinniġne secg by most editors of the poem, who have thus regarded fela as a scribal interpolation. As the editors of Klaeber IV have suggested, the rationale behind such an interpolation probably is that the scribe mistook 1378b, ðǣr þū findan miht, and 1379a as a verse pair, thereby introducing fela in order to remedy the lack of f alliteration. This is a credible hypothesis, since the initial consonant of the word interpolated by the scribe into 1379a alliterates with that of findan in 1378b. Why the scribe should have failed to recognize 1379b, sēċ ġif þū dyrre, as the pair of 1379a in the first place, however, remains an unresolved crux in Beowulf textual criticism. The present essay argues that because of his deficient understanding of the principles of verse construction, the scribe was unable to recognize the finite verb sēċ as the legitimate alliterative link of l. 1379a. By virtue of their intermediate prosodic status, the metrical behaviour of finite verbs, unlike that of stress-words and proclitics, is variable. Such variability is governed by a series of metrical rules whose ultimate rationale is the preservation of the four-position structure of the verse. If the scribe was ignorant of the four-position principle, as this essay argues, then he would have lacked the ability to scan the finite verb sēċ correctly. It is this inability to scan sēċ that would have led him to prefer the pairing of 1378b and 1379a, despite their failure to alliterate, to the pairing of 1379a and 1379b. Besides positing a new metrical rule that has not been fully articulated in previous scholarship, this essay explores the crucial implications that the scribes’ deficient understanding of Old English metre has for textual criticism.

Acknowledgements

I would like to thank R.D. Fulk and Leonard Neidorf for reading this essay in draft and offering invaluable suggestions.

Notes

¹ See Dobbie 1953. The translation is mine.

² Kemp Malone and Johannes Hoops have recommended editorial retention of fela (see Malone 1930; Hoops 1932: 166). The untenability of their positions will become apparent below.

³ Henceforth cited as Klaeber IV.

⁴ For two alternative, more conjectural explanations, see Pope 1952: 505–506; and Lehmann 1971: 38–39.

⁵ By functional alliteration is meant the alliteration of the line’s ictic positions, which binds the on- and the off-verse together. In this essay, non-functional alliteration refers to the ornamental alliteration of finite verbs that are placed in the clause-initial drop of an on-verse. This is the terminology adopted by A.J. Bliss in his discussion of the metrical value of finite verbs in Beowulf (see Bliss 1967: §§12–26).

⁶ In this essay, a distinction is maintained between metrical ictus and prosodic stress. While prosodic (also phrasal or sentence) stress is the emphasis that certain words receive within larger syntactic units in ordinary linguistic usage, ictus refers to the intensity that is accorded to some words in poetry for the sole purpose of preserving the integrity of the basic metrical structure of the verse. Metrical ictus and prosodic stress usually but not necessarily correlate. This distinction is especially important with regard to the metrical behaviour of finite verbs, as will be seen below. For more on the correlation between stress and ictus, see n. 14 below; see also Fulk 1992: §§255–268; 1998b: 239.

⁷ On linguistic compounding in the Germanic languages, see Krahe & Meid 1967: §15.

⁸ This idea was first advanced by Jerzy Kuryłowicz (1970: 16–20). For a detailed exposition of the metrical structure of the long line and its relation to functional alliteration, see Russom 1987: 71–73; 1998: 64–86; Terasawa 2011: 19–21; and Pascual 2014: 816. Daniel Donoghue (1990: 71) found this analysis of the structure of the Old English line problematic, because it conceptualizes the first ictic position of the off-verse, which has traditionally been considered the most prominent in the line, as weaker than the first ictic position of the on-verse. As will be seen further below, however, from the point of view of verse composition and reception, it is more natural to regard the first ictic position of the on-verse as the most prominent in the line (see Fujiwara 1990: 231; Terasawa 2011: 20 n. 6).

⁹ Klaeber’s 3rd edition of Beowulf is henceforth cited as Klaeber III.

¹⁰ See also Kendall 1982: 52; Fulk 1992: 58 n. 97; and Klaeber IV: 334 and n. 1. The reliability of Krackow’s law is confirmed by Russom’s observation that verses like Beowulf 3105a, bēagas ond brād gold, with a final ictic noun, systematically require double alliteration, while verses like 1569b, secg weorce ġefeh, with a final ictic finite verbal form, usually feature single alliteration (Russom 1987: 84–86, 96–97).

¹¹ Other fela-compounds in the poem are felafricgende (2106a), felaġẹōmor (2950a), felahrōr (27a), and felamōdiġra (1637a, 1888b). All of them participate in the functional alliteration of their lines.

¹² In the off-verse, the second ictic position is strictly forbidden to alliterate. The implication is that double alliteration is prohibited in the second half of the line (see Sievers 1893: §19; Bliss 1962: §15; Terasawa 2011: 4). The reason for this stricture against alliteration on the last ictus of the line is its low structural prominence. As will be seen below, this regularity must have been very useful for the real-time scansion of verses.

¹³ For the original formulation of this principle, see Sievers 1893: §8. For a full exposition of the four-position theory and its empirical sufficiency, see Cable 1974: 84–93; see also Pascual 2013. For a concise summary, see Stockwell & Minkova 1997: 67–68; Fulk 2002: 337–340; and Pascual 2014: 811–812.

¹⁴ According to the degree of ictus on a given syllable, three different metrical positions are distinguished: the lift, which accommodates a syllable that receives primary ictus; the half-lift, which contains a syllable with secondary or tertiary ictus; and the drop, which is paradigmatically formed by one non-ictic syllable. At the primary and secondary levels, metrical ictus correlates with prosodic stress. Thus, syllables with primary and secondary stress tend to occupy lifts and half-lifts, respectively. In the realization of primary and secondary ictus, prosodic stress is assisted by syllable length, to the extent that non-resolved short stressed syllables very rarely receive ictus. At the tertiary level, however, ictus is predicated solely upon syllable length, as Fulk has demonstrated (1992: §§255–268; see also Pope 2001: 140 n. 21).

¹⁵ Thus, the metrical principles of resolution and drop protraction, designed in order to enable the four-position rule, bring order to the apparently random fluctuations in the number of syllables in Old English verse. Resolution is a principle of phonological equivalence whereby a short stressed syllable and its immediate unstressed successor are made equivalent to a long stressed syllable in metrical terms (on syllabic length, see n. 16). Consequently, either a long stressed syllable or a resolved disyllabic sequence is able to occupy a primary or secondary ictic position. According to the principle of drop protraction, several unstressed syllables are able to occupy a single non-ictic position, or drop, as long as they are all adjacent and non-verse-final. For Sievers’s original formulation of resolution and drop protraction, see 1893: §§9–10. On the empirical reality of resolution, see Suzuki 1995; Pope 2001: 148–149; Fulk 2007c: 140; and Pascual 2014: 813–815.

¹⁶ A syllable is heavy or long if it contains either a long vowel or diphthong, or a short vowel or diphthong followed by one consonant. On Old English syllabification, see, for example, Hogg 2011: §§2.80–2.83. That the heavy middle syllable of trisyllabic words receives ictus is confirmed by the relatively high incidence of verses like Beowulf 1379a, sinniġne secg, and the non-occurrence of verses like the hypothetical *sinniġra secga, which must therefore be taken to consist of five positions. The palatal consonant /j/ represented by the letter <ġ> in words like sinniġne could be vocalized and absorbed into the immediately preceding vowel, thereby leaving a middle syllable containing a single vowel. It is important to note, however, that this sound change does not compromise the four-position interpretation of l. 1379a. Vocalisation took place mainly in West Saxon, not in the original Anglian dialect of Beowulf. But even if it had taken place, the resulting vowel would still have been long, /i:/, so that the length of the middle syllable would have remained unaffected. The four-position structure of the verse sinniġne secg is thus beyond question. On the vocalisation of postvocalic /j/, see Luick 1914–1940: §257; Campbell 1959: §§266–272; Brunner 1965: §126; Hogg 2011: §§7.69–7.71. On the orthographic evidence for the inception of this phonological change, see Colman 1983.

¹⁷ The German term for stress-word, originally proposed by Hans Kuhn, is Satzteil. The original German words for proclitic and particle, which are described further below, are Satzteilpartikel and Satzpartikel, respectively (see Kuhn 1933).

¹⁸ If they follow rather than precede the stress-words on which they depend, however, proclitics take ictus (see, for example, Bliss 1962: §10; Pope 2001: 137; Terasawa 2011: 28). For a more detailed account of the promotion of proclitics to ictic positions, see Lapidge 2006.

¹⁹ “Old and very grey, with his company of warriors.”

²⁰ They include finite verbs, pronouns, subordinating conjunctions, and many monosyllabic adverbs.

²¹ A verse clause is any clause that occurs in Old English poetry, regardless of the number of verses that accommodate it. (That number varies from several to less than one, since a single verse is able to accommodate two clauses, as in, say, l. 1379b). A verse clause typically consists of a series of stressed elements, their attending proclitics, and particles. Reduced to its minimum expression, however, a verse clause can consist of a single finite verb (see Bliss 1967: §§9–10).

²² For summaries of Kuhn’s first law, see Campbell 1970: 94; Lucas 1990: 294; Kendall 1991: 17–18; Hutcheson 1992: 129; Momma 1997: 56–64; Orton 1999: 289 n. 11; Pope 2001: 136-138; Terasawa 2011: 95–96; Fulk 2012a: 558–559; 2012b: 389. On the empirical sufficiency of Kuhn’s first law, see Donoghue 1997.

²³ “He who many [battles had survived] saw then in the wall.”

²⁴ “He also saw a standard all of gold hanging.”

²⁵ Notice that although swylċe is a disyllabic adverb, it is invariably non-ictic in Beowulf.

²⁶ The stricture against polysyllabic drops at the end of the verse is a reflection of the Indo-European principle of closure, which requires more fixed structures at the end of a metrical domain. Another reflection of such a principle is the regularity known as Fulk’s law, according to which the metrical value of disyllabic sequences with a short penultimate syllable under ictus at the tertiary level is determined by their position within the verse: in the onset they occupy a single position while in the coda they occupy two (for a definition of the onset and the coda of the verse, see n. 40). This regularity originated by analogy to Kaluza’s law, according to which two short syllables under ictus at the secondary level are usually resolved only in the onset of the verse. On the Indo-European principle of closure, also known as right justification, see Foley 1985: 12. On Fulk’s law, see Fulk 1992: §§221–245; and Pascual 2013: 58. On Kaluza’s law, see Kaluza 1896; 1909: 57–59; Fulk 1992: §§170–183; 1998a; 2007b: 317–323; Cable 1994; 2003; and Neidorf & Pascual 2014.

²⁷ “The victorious triumphant then saw.”

²⁸ If the finite verbal form ġeseah received ictus, then the adverb ðā, a particle, should also receive ictus, since it would be displaced from its unmarked verse-clausal position beside the prefix ġe-. Ictus on ðā, however, would generate the unprecedented six-position metrical structure x / \ / \ x (see Bliss 1967: §20).

²⁹ For the same interpretation of verse-clause-initial alliterating verbs, see Bliss 1967: §20; Kendall 1983: 7–9; 1991: 22–24.

³⁰ See n. 26 above.

³¹ This new metrical rule was partially adumbrated by Sievers (1893: §24.3), Bliss (1967: §15), and Kendall (1991: 23). It is defined as an inherited feature from the older heroic lays by Campbell (1962: 16–17). If Campbell’s position were correct, however, we should expect to find non-ictic alliterating finite verbs in passages that supposedly do not derive from ancient lays.

³² “The kinsman of Ælfhere; he saw his liege lord.”

³³ “The sad at heart [made a speech]; – he looked at the unloved – .” This line is quoted from Klaeber III, in which dashes are used to isolate the off-verse as a separate clause. The editors of Klaeber IV have deleted these dashes, probably to make the verse conform to Kuhn’s first law (notice, however, that the verse is still included in a list of half-lines that violate Kuhn’s first law in the appendix on textual criticism; see Klaeber IV: 334). According to the rule of off-verse-initial finite verbs formulated in this essay, ictus on seah is not an oddity, since the verb is able to alliterate with the on-verse.

³⁴ Thus, if the alliteration on finite verbs like ġeseah in on-verses like Ġeseah ðā siġehrēdiġ were accidental, we would expect to find approximately the same incidence of non-ictic alliterating finite verbs in the off-verse. There is a single off-verse in Beowulf that does not seem to conform to this rule: l. 1727b, hē āh ealra ġeweald. The finite verb āh is able to alliterate with the on-verse and yet it cannot be ictic, since ictus on āh would result in the metrical structure of a type E, which does not allow the addition of anacrustic syllables. The prosodic status of the finite forms of āgan was perhaps lower than that of other finite verbs, and hence its alliteration was accidental (see Bliss 1967: §18). If āh was particularly weak in prosodic terms, perhaps there was an elision between hē and āh, so that the initial vowel of āh did not interfere with the four-position scansion of the verse.

³⁵ Determining the applicability of this metrical rule in poems other than Beowulf will require close examination of the data and should form the basis for a subsequent study. It does not appear improbable, however, that, just like Kuhn’s first law, this rule regularly operates in classical Old English verse.

³⁶ It is also one of the few verses in Beowulf that contain a verse-internal clause boundary (see Kendall 1991: 89–90).

³⁷ On the stricture against verse-final polysyllabic drops, see n. 26. The clause-final location of dyrre probably favoured its ictic scansion, since that is a location where finite verbs tend to receive ictus.

³⁸ The scribe’s chronological distance from the composition of Beowulf is a probable factor here. Recent philological research has provided compelling reasons to believe that the composition of Beowulf antedated the production of its extant manuscript by approximately three centuries. See Lapidge 2000; Fulk 2007a; 2007b; Clark 2009; Neidorf 2013; Neidorf and Pascual 2014; and the fourteen essays recently published in Neidorf 2014.

³⁹ On this scribal habit, see, for example, Klaeber IV: cliv–clix; Fulk 1996; 1997.

⁴⁰ According to Fulk’s law, the middle syllables of mihtiga and mōdiga should be ictic, because they appear in the coda. The coda of the verse comprises the last full ictus and all subsequent syllables. The linguistic material preceding the coda is the verse onset (see Fulk 1992: 201 n. 60). On the syncope of middle vowels, also known as High Vowel Deletion, see Fulk 2010, which contains the most comprehensive and explanatory study of the exact application of this phonological change. See also Sievers 1898: §144 and n. 1; Wright & Wright 1925: §216; Kiparsky & O’Neil 1976: 534; Lass 1994: 101; and Fulk 2014: §54.

⁴¹ On the unmetricality of catalectic measures, see Sievers 1893: §§10.1, 180. See also Vetter 1872: 33; Cosijn 1894: 441 and n. 2; Trautmann 1907: 140; Pope 1966: 320–321, 334, 372; Bliss 1967: §84; Russom 1987: 117, 129; Fulk 1992: §209; 1996: 5–6; Hutcheson 1995: 156–158; Suzuki 1996: 120; Klaeber IV: 332; Terasawa 2009; 2011: 49–52; and Pascual 2013.

⁴² For this diagnosis, see, for example, Sisam 1953: 29–30, 44; Lapidge 1993; 1994; Fulk 1996; 1997.

⁴³ See Stanley 1984: 257.