Perfect and Imperfect Rhyme: Romances in the abab Tradition

STUDIES IN PHILOLOGY Volume 111 Winter 2014 Number 4 Perfect and Imperfect Rhyme: Romances in the abab Tradition by Judith A. Jefferson, Donka Minkova, and Ad Putter This article focuses on a group of Middle English romances composed in four-line stanzas rhyming abab. Surviving examples of this form include Thomas of Erceldoune, The Sowdone of Babylon, The Knight of Courtesy, and the fragmentary Partonope of Blois. Since these romances are from different dialects areas, the verse form appears to have been a popular one in medieval England. Examining the quality of the rhymes in the extant manuscripts, we show that both the original poets and the scribes of these romances were happy to tolerate imperfect rhyme. Two common types of imperfect rhyme, “feature rhyme” and “subsequence rhyme,” are discussed, and we provide analogues for such rhymes in medieval and modern song, from nursery rhymes to Latin hymns. We conclude by suggesting that the use of so-called “imperfect” rhyme is linked with oral performance and that it was in fact perfectly acceptable in this context. D ISCUSSION of rhyme in medieval verse, whether undertaken by editors as a means of establishing dialect or by philologists interested in phonemic changes, has always tended to concentrate on vowels rather than on consonants.¹ There are good reasons for this. As Roger Lass has observed, “Major systemic changes . . . are not ¹ Recent examples of such discussions focusing on Middle English romances are Ad Putter, Judith Jefferson and Donka Minkova, “Dialect, Rhyme, and Emendation in Sir Tristrem,” Journal of English and Germanic Philology 113 (2014): 73–92, Rhiannon Purdie, Anglicising Romance: Tail-Rhyme and Genre in Medieval English Literature (Cambridge: D. S. Brewer, 2008), 158–242, and Masa Ikegami, “Rhyme Evidence of the Great Vowel Shift in the Ashmole Sir Ferumbras (c.1380),” Nowele 30 (1997): 3–19. Ikegami uses rhyme to argue that the change of, e.g., Middle English e: to i: had already started in the fourteenth century. 631 © 2014 Studies in Philology, Incorporated Sheet 4 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 5 of 257 632 Romances in the abab Tradition prominent in the history of English consonants. Indeed, the consonant system as a whole has remained relatively stable since Old English times.”² This does not mean, of course, that consonants did not change at all,³ but it does mean that consonants are likely to provide much less evidence of either dialect or of phonemic change than is the case with vowels. Rhyme, however, involves consonants as well as vowels, and in the case of one particular type of rhyme—rhyme which is imperfect in respect of its consonants—the consonants are in fact of greater interest. E. G. Stanley has drawn attention to the widespread use in Middle English of rhymes where the vowels are identical but the consonants, by contrast, are simply close in sound,⁴ and in this essay we would like to pursue this topic further, by considering the use of rhymes that appear to be imperfect in respect of their consonants in one particular type of Middle English romance: romances rhyming abab. We shall also consider possible origins of this practice—particularly in Latin hymnody— and will place it in the context of oral poetry in general. The abab stanza form is not one that scholars today especially associate with Middle English romances, but it must have had some currency in the medieval period, for it is found in at least four romances, which will form the main focus of this investigation:⁵ 1. The Sowdone of Babylon. 3274 lines. S.xiv/xv.⁶ One copy: Princeton University Library MS Garrett 140. Ed. by Emil Hausknecht, EETS ES 38 (1881) and by ² Roger Lass, “Phonology and Morphology,” in The Cambridge History of the English Language, Volume 2: 1066–1476, ed. Norman Blake (Cambridge: Cambridge University Press, 1992), 57. This is, of course, a generalization. See, e.g., John Anderson, “A Major Restructuring of the English Consonant System: The De-linearization of [h] and the Deconsonantization of [w] and [j],” English Language and Linguistics 5 (2001): 199–213. ³ See e.g., Lass’s discussion of obstruents and sonorants in Lass, “Phonology and Morphology,” 57–67. ⁴ See E. G. Stanley, “Rhymes in English Medieval Verse: From Old English to Middle English,” in Medieval Studies Presented to George Kane, ed. Edward Donald Kennedy, Ronald Waldron, and Joseph C. Wittig (Cambridge: D. S. Brewer, 1988), 36, 49 and passim, and see also the discussions by Saara Nevanlinna, “Observations on the Loss of Final Plosive Consonants in Late Middle English Rhyme-words,” in Beowulf and Beyond, ed. Hans Sauer and Renate Bauer (Frankfurt am Main: Peter Lang, 2007), 271–92, and, for headrhyme, Donka Minkova, Alliteration and Sound Change in Early English (Cambridge: Cambridge University Press, 2003). ⁵ We propose to deal separately with septenaries such as those found in Sir Ferumbras (ll. 1–3410), which could be set out as quatrains (a⁴b³ a⁴b³). For the possible relationship of the 4343 stanza to the septenary, see Max Kaluza, A Short History of English Versification,trans. A. C. Dunstan (London: George Allen and Co. Ltd., 1911), §170. It seems equally possible, however, that the 4343 stanza was a variant of the 4-beat quatrain, with silent fourth beats in ll.2 and 4. ⁶ For the date, see Hausknecht, ed., xlvi, and Lupack, ed., note to l.46. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 6 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 633 Alan Lupack in Three Middle English Charlemagne Romances (Kalamazoo, MI: Medieval Institute Publications, Western Michigan University, 1990). 2. The Knight of Courtesy and the Lady of Faguell. 504 lines. Second half of s.xv.⁷ Extant in a quarto pamphlet published by William Copland in 1586, now in the Sheldon Collection of the Bodleian Library, Oxford. Reprints appear in Joseph Ritson, Ancient English Metrical Romances, 3 vols. (London: Bulmer and Co., 1802), 3.193–218; Francis James Child, English and Scottish Ballads, 8 vols. (Boston: Little Brown and Co., 1857), 1.188–210; William Carew Hazlitt, Remains of the Early Popular Poetry of England, 4 vols. (London: John Russell Smith, 1886), 2.55–87; and E. Goldsmid’s revised edition of Ritson, 3 vols. (Edinburgh: E. and G. Goldsmid, 1885), 3.172–88. Ed. by Elizabeth McCausland, “The Knight of Curtesy and the Fair Lady of Faguell”, Smith College Studies in Modern Languages 4 (1992), issued as an independent pamphlet. 3. The fragment of the shorter version of Partonope of Blois. 308 lines.⁸ Treatment of final –e suggests a date in the fifteenth century.⁹ Extant in Tokyo, Takamiya MS 32, olim Penrose, olim Delaware. Reprinted from the R. C. Nichols edition for the Roxburghe Club (London, 1873) by A. Trampe Bödtker, EETS ES 109 (1912, for 1911), 481–88. 4. Thomas of Erceldoune. 700 lines. Second half of s.xiv.¹⁰ Extant in five manuscripts: the Thornton MS, Lincoln Cathedral MS 91 (T);¹¹ Cambridge University Library MS Ff.5.48 (C); and three British library manuscripts: MS Cotton Vitellus E.X (Cot), MS Lansdowne 762 (L), and MS Sloane 2578 (S). Parallel text edition by James A. H. Murray, The Romance and Prophecies of Thomas of Erceldoune, EETS (1875); critical edition based on the Thornton MS by Alois Brandl, Thomas of Erceldoune (Berlin: Weidmannsche Buchhandlung, 1880); and, most recently, a second parallel text edition by Ingebord Nixon, Thomas ⁷ For the date, see McCausland, ed., vii and x–xi. ⁸ Boffey and Edwards state that this poem is in twelve-line stanzas, but, though the first three stanzas are linked by rhyme, this is not the case beyond this point. See Julia Boffey and A. S. G. Edwards, A New Index of Middle English Verse (London: The British Library, 2005), 272, item 4081. ⁹ See rhymes such as done v.pp.:sone adv.(193:195); fedde v.pp.:to bedde (201:203), both suggesting that such –es were not pronounced. ¹⁰ For the date, see Nixon, ed., 2.44–48. Dating is complicated because the actual prophecies may well have been added over a period of time but, as Nixon observes (2.45), “Fytte I must obviously have been composed after the lifetime of the historical Thomas of Erceldoune, at a time when his reputation as a prophet and a poet was established and widely known, that is, not earlier than the first half of the fourteenth century, and probably rather later.” If Laura Loomis is correct in her belief that Chaucer knew the romance when he wrote “Sir Thopas”, then, as Nixon says, this fitt must have been circulating some years before the turn of the century. See Loomis, “Sir Thopas,” in Sources and Analogues of Chaucer’s Canterbury Tales, ed. W. F. Bryan and G. Dempster (Chicago: University of Chicago Press, 1941), 516, and for further discussion of this topic, see Joanne A. Charbonneau, “Sir Thopas,” in Sources and Analogues of the Canterbury Tales, ed. Robert M Correale with the assistance of Mary Hamel, 2 vols. (Cambridge: D. S. Brewer, 2005), 2.649–714. ¹¹ This has a prologue rhyming ababababab, not present in the remaining manuscripts. Sheet 6 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 7 of 257 634 Romances in the abab Tradition of Erceldoune, 2 vols, (Copenhagen: Akademisk Forlag, 1980). There is also an edition (SS) printed in London in 1652 as part of Sundry Strange Prophecies of Merlin, Bede, Becket and Others, apparently drawn from a witness no longer extant but resembling that of the Sloane MS. This last has been reprinted and discussed by W. P. Albrecht in The Loathly Lady in “Thomas of Erceldoune”, University of New Mexico Publications in Language and Literature 11 (1954), 80–94.¹² Since these texts come from a number of different regions (The Sowdone from the East Midlands,¹³ The Knight of Courtesy from London,¹⁴ Partonope from Kent,¹⁵ and Thomas of Erceldoune from the North¹⁶), it is clear that the abab stanza was, in fact, a widely used form. As far as vowels are concerned, the rhymes in these poems appear to be accurate—as accurate as they are, say, in the poetry of Chaucer. This does not, of course, mean that apparently inaccurate rhymes are not present in the manuscripts, but there are normally particular explanations for these, such as lexical substitution or the replacement of authorial by scribal dialect forms. Several examples of such scribal substitutions are given by Hausknecht in his edition of the Sowdone, and he suggests a number of plausible emendations,¹⁷ e.g. of Agremare:there (33:35) to Agremore:thore (cf. rhymes at 126:128, 1375:1377, and for forms of “there” in the East Midlands, see LALME 1, Dot Maps 319, 321 and 322); and of oost:beste (463:465) to reste:beste (cf. rhymes at 3271:3273). Similarly Murray, in his edition of Thomas of Erceldoune suggests that the rhyme words found in T at 61:63, i.e Iral fine:schone should be emended to Iral stane:schane.¹⁸ If we leave aside cases likely to be scribal in origin, we can see that vowel variation in rhyme was unacceptable to the authors of these poems. ¹² For further discussion of this particular version, see Nixon, ed., 2.6 ¹³ See Hausknecht, ed., xxiv–xl. ¹⁴ McCausland, ed., vii. ¹⁵ See Angus McIntosh, M. L. Samuels, and Michael Benskin, A Linguistic Atlas of Late Medieval English (henceforth LALME), 4 vols. (Aberdeen: Aberdeen University Press, 1986), 1.202, source no. 5970. ¹⁶ Rhymes show that the poem comes from an area where OE /a:/ remains in Middle English, and it is therefore clearly Northern and possibly—though this is less likely— Scottish (see Nixon, ed., 2.9–18). ¹⁷ Hausknecht, ed., xliv and see also emendations suggested in the notes to ll. 1221, 1327, 1372, 1538, 1615, 1816, 2390 (where the accuracy of the resulting rhyme is dependent on changes associated with the Great Vowel Shift; see Ikegami, “Rhyme Evidence”), 2399, 2748, 2805, 2845, 2887 (though note that the emendation of the 2889 reading of assaye to of assyse=“customary” would also be possible), 2891. ¹⁸ See Murray, ed., lxxi. The emendation is necessary because of scribal misunderstanding of the first item (Iral appears to be a type of precious stone, but the word is very rare) and dialect translation of the second. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 8 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 635 What does seem possible, however, is that the abab poets, unlike poets such as Chaucer and Gower, were happy to make use of certain types of imperfect rhyme: rhymes which were imperfect in respect of their consonants. The most common of these are rhymes where the consonants share particular distinctive features but differ in others, a practice for which Arnold Zwicky has coined the term “feature rhyme.”¹⁹ There is one example, for instance, in the Sowdone of a rhyme which involves two different voiceless plosives, something which is common in various types of traditional verse (“Little Tommy Tucker, Sings for his supper” etc.): Rowlande met with Ferumbras And gafe him such a stroke That al astonyed þof he was, It made him lowe to stoupe.²⁰ More common is the practice of rhyming words with different nasals: A drift of wedire vs droffe to Rome, The Romaynes robbed vs anone; Of vs thai slowgh ful many one. With sorwe and care we be bygone. (Sowdone 76–79) Thee Messageris thanne wenten hoom, And tolden the Mayden this tidynge. Soo fayr a chyild sawe they never noon: Hee is Eerl of Bloys and cozyn to the kynge. (Partonope 69–72) “If I dyd nat, I were to blame.” Than sighed the lady with that worde; In dolour depe her hirte was tane And sore wounded as wyth a sworde. (Knight of Curtesy 161–4) ¹⁹ See A. M. Zwicky, “Well, this Rock and Roll has got to stop. Junior’s head is hard as a Rock,” in Papers from the Twelfth Meeting of the Chicago Linguistics Society, ed. S. Mufwene, C. Walker, and S. Steever (Chicago: University of Chicago Press, 1976), 677. ²⁰ It seems likely that both the vowels here would have been pronounced as /u:/. As the OED notes, stoop can be compared to a coop and droop, all cases where ME ū before p has remained unchanged instead of becoming /aυ/, while Dobson provides evidence that forms in /u:/ were possible for the past tense form of the verb “to strike” with prior development of the vowel to close ō (cp. the modern form “struck” with shortening), whence possible analogical forms of the noun. See E. J. Dobson, English Pronunciation: 1500–1700 (Oxford: Clarendon Press, 1957), §151. For discussion of the rhyme strook (n.) with sook (=“sucked”) in Hoccleve’s Regement of Princes, see Masa T. Ikegami, Rhyme and Pronunciation (Tokyo: Hogaku-Kenkyu-Kai, 1984), 104–5 and 110, n.6. Sheet 8 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 9 of 257 636 Romances in the abab Tradition Further examples of this type of rhyme can be found at, e.g., Sowdone 177:179 and 2552:2554 (hym:wynne), 364:366 (Rome:done), 447:449 and 859:861 (him:kyn), 452:454 (came:man),²¹ 747:749 (Sowdan:some), 1323:1325 (came:than), 1488:1490 (by- came:ranne), 2004:2006 (hym:skyne), 2028:2030 (tyme:pyne), 2443:2445 (done:come), 2351:2353 (inne:him), 2999:3001 (withynne:him); 3036:3038 (dam:than); Knight of Curtesy 217:219 (came:wan), 393:395 (ran:came).²² Also common, are instances where words in <ng> rhyme with words in <n> or <m>. In the case of a rhyme such as come:yonge (Knight of Curtesy 134–6), the imperfection of the rhyme is unarguable: Than let he do crye a feest For euery man that thider wolde come, For euery man, bothe moost and leest; Thyder came lordes, bothe olde and yonge. It is more difficult to be sure of the status of the <n>:<ng> rhymes, and Nevanlinna sees these as examples of final stop deletion (i.e. as perfect rhymes).²³ In fact, however, the majority of Nevanlinna’s examples involve weak –ing endings, and it seems likely that there is a distinction to be drawn between rhymes of this type²⁴ and those where the syllable ending in <ng> occurs in a monosyllabic word or in a syllable bearing the main word stress. As Zwicky observes, the process /ŋ/ > /n/ is far less likely where the syllables involved are stressed syllables,²⁵ and, al²¹ The vowel in cam is presumably short, as in, for example, the rhyme cam:ram In Chaucer’s description of the Miller in the “General Prologue” (ll.547–8). Such forms were clearly available to ME poets beside forms in close ō from OE c(w) ōm and it seems likely that forms in short –a- were present (though unrecorded) in OE; see Roger Lass, Old English: A Historical Linguistic Companion (Cambridge: Cambridge University Press, 1994), 157, and for the replacement of short by the corresponding long vowels in the preterites of strong verbs (which, with later shortening, explains the NE vowel), see Sherrylyn Branchaw, “Survival of the Strongest: Strong Verbs in the History of English,”, unpublished PhD dissertation (UCLA, 2010), 100–34 (she does not, however, deal specifically with “come”). ²² Given the other examples, it seems unlikely that the use of the pronoun “him” to rhyme with words ending in /n/ in the Sowdone reflects a pronunciation in /n/. LALME, in any case, records such forms almost exclusively in the South and West (see LALME 4, item 152 and p.314), with further possible instances in Southern Suffolk and in Norfolk near the Wash, but, since the Sowdone shows no other East Anglian characteristics, this seems unlikely to be relevant. ²³ Nevanlinna, “Observations,” 277–78. ²⁴ As at Sowdone 84:86 (tyþinge:appolyne), 2347:2349 (Mapyne:endinge), 3171:3173 (serpentyne:endyng), and see also 948:949 (engyne:offerynge; though something has gone wrong with the abab pattern here), 2045:2047 (lym:rennynge; though even with such reduction this is clearly an imperfect rhyme). ²⁵ Zwicky, “Well, this Rock and Roll,” 687. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 10 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 637 though, as Dobson points out, Jordan does provide evidence of –in for the weak ending –ing at a comparatively early date (i.e. before 1350),²⁶ neither provides evidence of a similar change in non-weak syllables.²⁷ Instances from our corpus where –<ng> occurring in a stressed syllable or in a monosyllabic word rhymes with –<n> include everychone:among (Sowdone 1972:1974), distruccion:wronge (Sowdone 959:61), kynge:Genelyne (Sowdone 3247:3249), and these should probably be defined as feature rhymes (/n/ rhyming with /ŋ/). These cases of <n>:<ng> and <m>:<ng> rhymes bring us to another type of imperfect rhyme, designated by Zwicky as “subsequence rhyme”,²⁸ where one of the rhymes has a subsequent phoneme or syllable not present in the other. Examples of subsequence rhyme are certainly present in our corpus; in these examples the final unshared element is normally a plural inflexion: Take your sporte, and kith your knyghtes,²⁹ Whan ye shalle haue to done; For to morowe, when the day is light, Ye mooste to the wallis goon. (Sowdone 2087–90) Ardeneys was a wyilde forest That no man durste huntte thare For liowns, liberdys, and other wylde beestis That gryisly were in holtis hare. (Partonope 109–112) ²⁶ Dobson, English Pronunciation, §399, and Richard Jordan, Handbook of Middle English Grammar, trans. and rev. E. J. Crook (The Hague: Mouton, 1974), § 175. ²⁷ This may suggest that the acceptability or otherwise of deletion may be related to meaning, since the distinction between /ŋ/ and /n/ in stressed syllables is often phonemic (thin vs. thing, gone vs. gong, sin vs. sing, ban vs. bang etc.). As far as the weak –ing ending is concerned, our findings do not support the argument put forward by Labov that –ing endings which would have had –ind in EME (i.e. participles) are more likely to be pronounced –in’ even in Present Day English than forms which had ME –ing (i.e. gerunds). See William Labov, “The Child as Linguistic Historian,” Language Variation and Change 1 (1989): 87. In our corpus, gerunds seem equally—if not more—likely to rhyme on /n/; see note 24 above, where only one of the examples cited does not involve a gerund. ²⁸ Zwicky, “Well, this Rock and Roll,” 677. ²⁹ The plural morpheme at the end of knyghtes (which results from an expansion) should perhaps simply be –s (giving knyghts). For loss of syllabicity in this and other inflections, see Christina M. Fitzgerald, “‘Ubbe dubbede him to knith’: The Scansion of Havelok and ME –es, –ed, and –ede ,” in Studies in the History of the English Language IV: Empirical and Analytical Advances in the Study of English Language Change, ed. Susan M. Fitzmaurice (Berlin: Mounton de Gruyter, 2008), 187–204. This does not, of course, mean that such cases do not involve subsequence rhyme. Sheet 10 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 11 of 257 638 Romances in the abab Tradition Some instances of this pattern involve plurals which follow a numeral, and, since such plurals were often uninflected, these particular examples are not entirely convincing.³⁰ There are, however, plenty of examples, like those quoted above, where this is not the case,³¹ and, since there is little historical evidence for the loss of the plural morpheme except after numerals,³² these seem likely to be instances of subsequence rhyme. In spite of the arguments against the overestimation of functionalism presented by William Labov, it is tempting to view this pattern of loss and retention of plural /s/ in the English language in general as being related to the degree of meaning carried by the morpheme, since in cases where a numeral is present the plural morpheme is clearly redundant.³³ The one remaining possible type of imperfect rhyme involves the rhyming of words in /nd/ with words in /n/, as in the following: He lai cryande at the grounde Like a develle of Helle; Through the Cite wente the sowne, So lowde than gan he yelle. (Sowdone, 435:437) This is not a straightforward example, however, since the form “sound” was perfectly possible by the late fourteenth century, as was the loss of ³⁰ See Tauno F. Mustanoja, A Middle English Syntax. I: Parts of Speech (Helsinki: Société Néophilologuique, 1960), 57–58, and for such rhymes see, e.g., Sowdone 68:70 and 1455:1457. ³¹ See, e.g., the following in the Sowdone: tidyngys:thinge (796:797), Walles:alle (2463:2466), nede:stedes (2680:2682), moone:stoones (944:946), A further examples of feature rhyme (t:d) plus subsequence rhyme (Ø:–es) occurs at Sowdone 2196:2198 viz. Ascopartes:liberde. ³² Evidence from modern English dialects suggests that plural /s/ will normally be retained even where other types of /s/ are lost. See, e.g., Labov’s discussion of African American practice in William Labov, “The Study of Language in Its Social Context” in Labov, Sociolinguistic patterns (Philadelphia: University of Pennsylvania Press, 1972), 221. The Middle English pattern (zero plural after a numeral only) occurs in African-American dialects; see Lisa Green, African American English: A Linguistic Introduction (Cambridge: Cambridge University Press, 2002), 208 and 253. ³³ See Mustanoja, Syntax, 58. Whether patterns of deletion are related to the desire to retain meaning is a matter of some controversy. Paul Kiparsky has argued that the relationship of d/t deletion in English to morphemic structure does demonstrate such a desire, but Labov notes that similar losses often occur where they do affect meaning, while Tagliamonte and Temple argue that, for British English at least, morphological class does not significantly affect t/d variation. See Kiparsky, Explanation in Phonology (Dordrecht: Foris, 1982), 89; Labov, “The Overestimation of Functionalism,” in Principles of Linguistic Change,Volume I: Internal Factors (Oxford: Blackwell, 1994), 547–569; Sali Tagliamonte and Rosalind Temple, “New Perspectives on an ol’ Variable: (t,d) in British English,” Language Variation and Change 17 (2005): 281–302. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 12 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 639 /d/ after /n/ in “ground”.³⁴ Given the number of rhymes which are imperfect in respect of their consonants elsewhere in the poem, it is simply impossible to judge whether this particular rhyme was intended by the poet to be perfect or not.³⁵ It seems clear, in any case, that the abab poems did permit certain types of imperfect rhyme, both feature rhyme and subsequence rhyme, even if the categorization of individual instances is sometimes a matter for some doubt. Thomas of Erceldoune is of particular interest here, for unlike the other poems in our corpus it is extant in more than one manuscript.³⁶ Its textual descent basically shows a bipartite stemma, with T, S, SS, and Ca on the one side and Co and L on the other.³⁷ Variation between the manuscripts often makes it possible to obtain a perfect rhyme by picking and choosing amongst the manuscripts, but there is evidence to suggest that partial rhymes involving nasals, stops etc. were, at the ³⁴ See OED s.v. sound n.³, and for the assimilation of d/t after dentals, see Dobson, English Pronunciation, §398, and Jordan, Handbook, §199. As David Harris observes, the loss of /d/ after /n/ is reflected in the pronunciation of modern words such as lawn < OF launde and woodbine< OE wudubind. See David P. Harris, “The Phonemic Patterning of the Initial and Final Consonant Clusters of English from Late Old English to the Present,” unpublished PhD dissertation (University of Michigan, 1954), 114, and note also ME scande~scanne v. (=“scan”) <Latin scandere, and hypercorrect pound v. < OE punian, and ancient < OF ancien. ³⁵ Nevanlinna who, with the support of rhyme evidence, supplies a series of examples of what she views as the loss of final stops, argues that ME verse fulfils Penzl’s requirement that only verse with predominantly pure rhymes can reveal phonemic change (see Nevanlinna, “Observations,” 272). However, her corpus includes a high proportion of verse which cannot be expected to be pure in respect of its consonantal rhymes. Thus Robert of Brunne’s Handlyng Synne (ed. Idelle Sullens, Medieval & Renaissance texts & studies 14 [Binghamton, NY, 1983]), which Nevanlinna uses for evidence of reverse spelling following the loss of final stops after fricatives, contains a variety of types of rhyme imperfect in respect of their consonants, e.g., clene: bapteme (199:200), com:gon (1804:5), flore: furþe (2233:4), while a number of her examples, including some of apparent losses of –d after n, are unconvincing for other reasons. In two cases, the rhymes she cites involve verbs adopted into English from Latin or French past participles (commawnd pret.: land, and accept pp.: kepte; “Observations,” 274), verbs which were initially adopted into English with preterites and past participles of this form. See Ole R. Reuter, On the Development of English Verbs from Latin and French Past Participles, Commentationes humanarum litterarum V.1.6 (Helsingfors, 1934), and for a summary of Reuter’s findings, see 126–45. Two further examples involve what Nevanlinna sees as the loss of past participle /d/ after /n/ following earlier syncopation of the vowel (ywonyd:none:sone:done and inclind:pyne:thine; “Observations,” 277), but in both these cases the original form may well have been an adjective (iwone< OE gewuna, enclin< OF enclin). ³⁶ See p. 00 above. ³⁷ Further groupings have been suggested, viz S and Ca (Brandl, ed., Thomas of Erceldoune, 4–6), T and Ca (Albrecht, The Loathly Lady, 76–77). There is also evidence of contamination across the bipartite stemma; see Nixon, ed., vol. 2, 6. For sigla used here, see footnote 11 above. Sheet 12 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 13 of 257 640 Romances in the abab Tradition very least, acceptable to particular scribes. See, for example, the rhyme cock:note which appears in SS (i.e. the sixteenth century edition) at 29:31: I heard the Jay, the Thristle cock, The Mavis minding of her song, The Woodwal with her merry note, That all the wood about me rung.³⁸ The Sloane MS (S), to which SS is most closely related, has no witness here, but the readings of the Thornton MS (T), the Cambridge MS (Ca), and the Lansdown and Cotton MSS (L and Co) are as follows: I herde þe jaye & þe throstyll cokke, The Mawys menyde hir of hir songe, The wodewale beryde als a belle, That alle þe wode abowte me ronge. (T) I sawe þe throstyl & þe iay, Þe mawes movyde of hyr songe Þe wodwale sange notes gay, Þat all þe wod a boute range. (Co; L slightly different but with the same rhymes) I herde þe iay & þe throstell, Þe mavys menyd in hir song Þe wodewale farde as a bell, Þat þe wode aboute me rong. (Ca) The reading cokke is treated by editors as an error, with Albrecht—who comments that it spoils the rhyme—offering it as possible evidence of a relationship between T and SS.³⁹ As Albrecht acknowledges, however, even if this is assumed to be the case, it is difficult to account for the manuscript readings, and it is perhaps worth at least entertaining the possibility that the SS reading is correct. It then seems possible that “bell” in T may have been suggested by “rung” in the following line, which would in turn be likely to lead to the Ca reading, and that the Co and L reading is an attempt to deal with what is considered to be an unacceptable rhyme (cock:note).⁴⁰ In any case, it is clear that whoever was ³⁸ Quotations are taken from Nixon’s edition. ³⁹ Albrecht, The Loathly Lady, 78. ⁴⁰ The consonants might not be the only problem here. There are also possible objec- tions to the vowel sounds (rhyming of long and short o), although long o in “cock” would Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 14 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 641 responsible for the witness from which SS is derived had no objection to rhymes involving different voiceless stops. Likewise, the Co scribe is clearly happy with rhymes of a similar type, although in this case the weight of manuscript evidence is against him (the T L Ca reading is given first, in order to make the context clear, since the Co reading is partly illegible): She led hym into a fay(re) (h)erbere,⁴¹ Þer frute groande was gret plente; Payres and appuls, bothe ripe þei were, Þe darte, and also þe Damsyn tre. (Ca, and similarly L and T, 177–180) . . . . . . . . . . . . . . .she led hym tyte, Þer was fruyte gret plente; . . . . . . . . . . . . . . .es þer were rype. Þe date & þe damese. (Co) The Co scribe is also clearly happy with /n/:/m/ rhymes. See for instance Co’s reading at 276:278 (once again, not all the Co material is legible):⁴² ‘Do buske the Thomas, þe buse agayne, ffor þou may here no lengare be. Hye the faste, with myghte & mayne, I sall the brynge till Eldone tree. (T, and similarly L Ca SS) ‘buske þe thomas, þou most a-gayn, here þou may no la. . . . . . . . . . . . . . . . hy þe $erne, at þou wer at hame, I sall þe brynge to.’ In this case too, the weight of manuscript evidence is against Co, but, given that T, Ca, and SS all appear on the same side of the stemma, and that there is clear evidence for contamination involving T and L,⁴³ it is certainly possible that the Co reading might be archetypal. In each of the above cases—and perhaps particularly in the first, where the Norse certainly have been possible in Scottish and Northern Dialects (see Dobson, English Pronunciation, §53, note 2). ⁴¹ In Nixon’s edition, round brackets indicate that the letters enclosed are almost illegible. ⁴² The medial vowel in “again” in this second example may well have been pronounced /a:/ in the area where Thomas was written (see Jordan, Handbook, §132), so that the rhyme agayn:hame would be exact as far as the vowel was concerned. ⁴³ See Nixon, ed., 2.6. Sheet 14 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 15 of 257 Romances in the abab Tradition 642 word tyte might be expected to cause problems⁴⁴—the Co readings are arguably the more difficult. Feature rhyme also appears to have been tolerated by other scribes. Rhymes involving /m/ and /n/ were clearly acceptable, for instance, in the L tradition: And crowne hyme at a towne of Scone, fforsothe vpon a Setterday. Bornes blode shall wend to Rome, To get lyve of the pope, yf they may. (L 425–8) S reads as follows: & crowned at þe towne of scoune on a serteine solemne daye. birdes bolde, bothe olde & yonge, shall to him drawe without naye. Ca, as far as it is possible to tell, appears to have shared the S reading, and SS, although it reads “stone” for “Scone,” is clearly descended from it. Feature rhyme is thus clearly indicated by the manuscript tradition. Whether the original rhyme was Scone and Rome or scoune and yonge is more difficult to say. Since T and Co read $onge and alde at 427 (a clear error for alde and $onge), the weight of manuscript evidence is against L, but even so L’s reading cannot be entirely discounted. It is the durior lectio, referring as it does to a precise historical circumstance, the fact that a special bull was obtained from Rome (from Pope John XXII) to allow the consecration of the boy king David II in 1331,⁴⁵ and it is worth noting that elsewhere, when L provides more detailed information, this seems likely to correspond to the original reading. Thus, the L reading at l.467 (where L alone names Robert II, grandson of Bruce and the first Stewart) forms part of a passage where, as Albrecht observes, “only L makes a clear, historically accurate, and properly rhymed transition.”⁴⁶ ⁴⁴ Such Northern words do cause difficulties for scribes elsewhere; see, for example, L trowe for T and Ca hope (in the sense of “fear”) l.78, L and Ca heuen, Co hens for T hethyne, l.294. ⁴⁵ “In 1331 on 24 November David king of Scots, the son of Robert de Bruce, was anointed king and crowned at Scone by Sir James Ben bishop of St Andrews, who had been specially authorized to do this by a bull of the most holy father John XXII. Before him no king of Scotland is said to have been anointed and crowned with this kind of ceremony.” See Scotichronicon, ed. and trans. Walter Bower, vol. 7 (Aberdeen: Aberdeen University Press, 1996), 71, and see also Brandl, ed., note to this line. ⁴⁶ Albrecht, The Loathly Lady, 74, and see Nixon, ed., 2.71, note to ll.461–468. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 16 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 643 Subsequence rhyme appears to cause rather more concern among the scribes, though the Ca scribe seems to be willing to accept it: Rachis lay lappand on þe dere blode, Þe cokys þei (stode with) dressyng knyves, brytnand þe dere as þei were wode; Reuell was among þem rife. (Ca, ll.265–8) Here we find the common pattern of a plural inflection present in one rhyme word but not in another. It is true that T and SS both read “knife” at this point (Co’s rhyme words are lost), but the plural is present on both sides of the stemma (i.e. in L as well as Ca), though the reading in L suggests that some scribe in the L tradition may perhaps have felt unhappy about it, even if his replacement does not necessarily improve things:⁴⁷ Raches lay lapyng of his blode And kokes with dressyng knywys a hande Trytlege the dere as they were wode; There was ryfe reuoll a-monge. (L) Scribal variation at 314 may also suggest a certain unease about subsequence rhyme: To harpe or carpe, whare so þou gose, (þou] I L, $e Ca, gose] gon Co L Ca, can S SS) Thomas þou sall hafe þe chose sothely.’ And he saide, harpynge kepe I none, ffor tonge is chefe of mynstralsye.’ A further variation is provided here by Brandl, who emends to gonge; but it seems likely that the original “go” verb was a singular subjunctive, either ga or gange (both rhyme words probably originally had medial <a>), neither of which would provide an exact rhyme. In none of the cases cited above is the version with the imperfect rhyme the obvious, uncontroversial original reading, but nevertheless the evidence from the abab texts as a whole suggests that we should not automatically discount such possibilities; at the very least, they appear to have been acceptable to a number of scribes, no doubt because such ⁴⁷ It is possible, however, that the original reading here may have been han/hon without the <d>. See MED s.v. hōnd(e (n.), and for the loss of /d/ after /n/ see p.00 above. This would result not in subsequence rhyme but in a type of feature rhyme, i.e. /n/ rhyming with /ŋ/ (see p.00 above). Sheet 16 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 17 of 257 Romances in the abab Tradition 644 rhymes were familiar to them from other poems. Why such rhymes should occur in this particular type of poetry is another question, and one that can be approached both by comparison with later verse of this particular type and by consideration of the origins of such verse forms. To begin with the first: the associations of four-beat lines in four line stanzas, what Attridge calls the “four by four formation,”⁴⁸ have been extremely well documented. As Robbins Burling has shown, such stanzas are common in children’s verse, not only in English but in other languages, often with the abab pattern:⁴⁹ Old King Cole was a merry old soul, And a merry old soul was he; He called for his pipe and he called for his bowl, And he called for his fiddlers three.⁵⁰ Nor is this form confined to children’s verse. As Attridge observes, the four by four formation is “the most common of all the possible rhythmic patterns” and “is the basis of most modern popular music, including rock and rap, of most folk, broadside and industrial ballads from the Middle Ages to the twentieth century, of most hymns, of most nursery rhymes, and of a great deal of printed poetry.”⁵¹ Although Attridge ends this passage with a reference to printed works,⁵² most of the types of poetry he refers to in this list are in fact spoken or sung, and this is true too of a high proportion of the abab quatrains listed in the Index to Middle English Verse. The form is regularly used for material for speak⁴⁸ Derek Attridge, Poetic Rhythm: An Introduction (Cambridge: Cambridge University Press, 1995), 53–4. As Attridge observes elsewhere, this structure can include lines where the final beat is not realized, as in “Mary had a little lamb,” where the second and fourth lines have, strictly speaking, only three beats, but where it is necessary to follow each of these lines when reciting with a silent beat, and it may be worth bearing this in mind when considering these verse forms as they are used in Middle English. See Attridge, The Rhythms of English Poetry (London: Longman, 1982), 86–7. ⁴⁹ Burling, “The Metrics of Children’s Verse: A Cross-Linguistic Study,” American Anthropologist, n.s., 68 (1966): 1418–1441. ⁵⁰ Quoted by Burling, “Metrics,” 1420, and see also rhymes in Benkulu (“Dari mano punai melajang?/Dari kaju turun kepadi/ Dari man kasih sajang?/ Dari mato turun kehati,” i.e. “From where does the dove glide? / From the tree down to the padi / From where does love move? / From the eye down to the heart”; p.1431), ⁵¹ Attridge, Poetic Rhythm, 53–4. For the use of the four by four formation in ballads, see e.g., the following: “He took the halter frae his hose, / And of his purpose did na fail; / He slipt it oer the Wanton’s nose, / and tied it to his gray mare’s tail,” from “The Lochmaben Harper”, quoted in Attridge, Rhythms, 84. ⁵² Attridge cites, for instance, Wordsworth’s poem, “She dwelt among the untrodden ways”; see Attridge, Rhythms, 87, and for use of the form by Theodore Roethke, see ibid., 92. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 18 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 645 ing or singing: for carols, for instance, in the sense of songs in general. For an example with religious content (a “carol” in the modern sense), see, e.g., the following by John Audelay: Ther is a babe born of a may In saluacion of vs; That he be heryd in this day, Vene, Creatore Spiritus.⁵³ and for a more secular song, see the following from “In Praise of Brunettes”: god saue ale hem þat buþ broune, for þey buth trew as any stele; god kepe hem boþe in feld & toune, & þanne schal y be kept ful wel.⁵⁴ This form, then, is associated with oral delivery, and it is not surprising that abab is a very common stanza form in the small corpus Middle English songs that have survived with music.⁵⁵ It seems, moreover, likely that rhymes which are imperfect in respect of their consonants were particularly prone to occur in oral material rather than in material which was intended to be read. Such rhymes are still common, for instance, in modern popular verse, such as rock lyrics,⁵⁶ whereas they are, as Astrid Holtman observes, minimal in what she describes as “serious poetry.”⁵⁷ Indeed, the types of feature and subsequence rhyme which appear in the abab poems are exactly the types which are most likely to occur in such modern oral poetry. As Holtman notes, the two most important feature rhymes in the rock lyrics studied by Zwicky are rhymes involving different types of nasals [m:n] and [n:ŋ] (these make up 48% of the examples in his corpus),⁵⁸ as, for example, in the following: ⁵³ From BL MS Additional 40116 (C3), printed in Richard Leighton Green, ed., The Early English Carols (Oxford: Clarendon Press, 1935), 77–78. ⁵⁴ From Cambridge, Gonville and Caius College MS 383/603, 190, printed in Secular Lyrics of the XIV and XV Centuries, ed. Rossell Hope Robbins, 2nd ed. (Oxford: Clarendon, 1955), 30–1; Celia Sisam and Kenneth Sisam, eds., The Oxford Book of Medieval English Verse (Oxford: Clarendon, 1970), 455–6; Maxwell Sidney Luria and Richard L. Hoffman, eds., Middle English Lyrics (New York: Norton, 1974), 129. ⁵⁵ See E. J. Dobson and F. Ll. Harrison, eds., Medieval English Songs (London and Boston: Faber and Faber, 1979), items 16a, 18, 20, 22, 25, 30, 31, and 32. ⁵⁶ See, e.g., discussions by Zwicky, “Well, this Rock and Roll,” passim, and Holtman, A Generative Theory of Rhyme: An Optimality Approach (Utrecht: OTS dissertation series, 1966). ⁵⁷ Holtman, A Generative Theory, 198. ⁵⁸ Holtman, A Generative Theory, 227, Zwicky, “Well, this Rock and Roll,” 685. Sheet 18 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 19 of 257 Romances in the abab Tradition 646 Me and my gal, my gal, son, We got met with a tear gas bomb, I don’t even know why we come, Goin’ back where we come from. (Bob Dylan, “Oxford Town”)⁵⁹ We gotta get out while we’re young Cause tramps like us, baby we were born to run. (Bruce Springsteen, “Born to Run”)⁶⁰ Rhymes involving consonants which differ only in their place of articulation (as in the /p/:/k/ rhyme from The Sowdone quoted above on p.00) are also common in popular sung and spoken verse: Hush-a-bye baby, on the tree top When the wind blows, the cradle will rock (traditional) The things that sit and wait for you To stumble in the dark Will take the cobwebs from your eyes And plant them in your heart. (Joe Byrd, “The Elephant at the Door”)⁶¹ Similar continuities can be found in the types of subsequence rhyme present in Middle English and modern oral poetry. As is the case in the Sowdone, in Holtman’s corpus of Lou Reed songs, the most frequent subsequent rhymes are those involving [+s]: I dreamed I was president of these United States I dreamed I replaced ignorance, stupidity and hate. (Lou Reed, “The Day John Kennedy Died”) Dark party bars, Shiny Cadillac cars and the people on the subways and trains looking gray in the rain. (Lou Reed, “After Hours”)⁶² The fact that spoken material shows more imperfections in consonantal identity than in vocalic identity while the opposite is true of poetry which is designed to be read may suggest something about the differ⁵⁹ Quoted by Zwicky, “Well, this Rock and Roll,” 678. ⁶⁰ Quoted by Zwicky, “Well, this Rock and Roll,” 687. ⁶¹ Quoted by Zwicky, “Well, this Rock and Roll,” 685. ⁶² Between Thought and Expression: Selected Lyrics of Lou Reed (London: Viking, 1992), 78 and 24.[AU: PROVIDE AUTHOR’S NAME HERE] Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 20 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 647 ent modes of perception involved in reading and listening. As Holtman observes,⁶³ “when we listen to a song, the vowel quality of the rhyme word is usually of crucial importance. It is the sound that attracts most attention [because, as Holtman adds in a footnote, it is the most sonorous element of the syllable] and it is the sound that one expects to hear in the rhyme word in the next line.” Although Holtman is talking about song, where, as she observes, the vowel sound of the rhyme word can be lengthened on a musical note, it seems likely that the same is true to some extent of oral poetry in general: i.e., what the listener hears most clearly is the vowel. Since early verse was usually intended to be spoken, it therefore seems likely that such inexact rhymes were present in the very earliest rhyming poetry. Henry Lanz’s first rule of rhyme, for instance, reads as follows: No matter through what influences and at what time rime appears in the poetic literature of a given country, it always proceeds from more or less imperfect forms of terminal assonance to a more accurate repetition of sound.⁶⁴ It seems probable, then, that the author of the Sowdone and some of the other abab poems did indeed employ both feature and subsequence rhyme, either because these poems were themselves intended for oral delivery or because the stanza form was traditionally associated with such delivery.⁶⁵ Various models of imperfect rhyme were available to Middle English poets. Rhymes which are imperfect in respect of their consonants can be found in Welsh and Irish verse,⁶⁶ while forms of assonance are also ⁶³ Holtman, A Generative Theory, 197. ⁶⁴ Lanz, The Physical Basis of Rime (Stanford: Stanford University Press, 1931), 131–132. ⁶⁵ It may be that another reason for the acceptability of imperfect rhymes in the abab stanzas is that the rhyming syllables are further away from one another than they would be in couplets, though note the list of imperfect rhymes in the Owl and the Nightingale provided by Stanley; see E. G. Stanley, ed., The Owl and the Nightingale (London: Thomas Nelson and Sons, 1960), 111–12. Stanley does provide suggested alternative readings for many of these. ⁶⁶ For Welsh verse, see Thomas Parry, A History of Welsh Literature, trans. H. Idris Bell (Oxford: Clarendon Press, 1955), 25; and A. T. E. Matonis, “The Harley Lyrics: English and Welsh Convergences,” Modern Philology 86 (1988): 13. §IV.1, and 15. For Irish verse, see Matonis, “An Investigation of Celtic Influences on MS Harley 2253,” Modern Philology 70 (1972): 91–108. Matonis’s suggestion that the Harley Lyrics were subject to Irish influence has, however, been refuted by Fulton; see Helen Fulton, “The Theory of Celtic Influence on the Harley Lyrics,” Modern Philology 82 (1985): 239–254, esp. 243–246, and for the view that imperfect consonant rhymes in Irish cannot be viewed as “feature rhyme”, see Janet Grizenhout and Astrid Holtman, “Optimality and Poetic Rhyme in Early Irish,” in OTS Yearbook 1994, ed. Jan Don, Bert Schouten, and Win Zonneveld (Utrecht: Utrecht Institute of Linguistics), 43–62. Sheet 20 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 21 of 257 648 Romances in the abab Tradition to be found in Anglo-Latin hymns both from the Anglo-Saxon and the Middle English periods, as in the following examples: Summa nobis conferat in deitate trinitas in qua gloriatur unus per cuncta secua deus.⁶⁷ Kyrie cunctipotens; A superna sede Vota suscipe clemens, Nobis miserere.⁶⁸ Ostensa sibi vulnera In Christi carne fulgida Resurrexisse Dominum Voce fatentur publica.⁶⁹ In fact, while Welsh is clearly a possible source for English rhymed verse, especially perhaps on the Welsh borders (Matonis suggests, for instance that the Harley lyric “Annot and John” may well have been influenced by Welsh verse forms),⁷⁰ in the country as a whole, a source in Latin verse may seem more likely, if only because of the wider reach and greater familiarity of the latter.⁷¹ Michael McKie, for instance, ar⁶⁷ See Inge B. Milfull, ed., The Hymns of the Anglo-Saxon Church: A Study and Edition of the Durham Hymnal (Cambridge: Cambridge University Press, 1996), item 60, p.251. ⁶⁸ Analecta Hymnica medii aevi, ed. Guido Maria Dreves, Clemens Blume, and Henry Marriott Bannister, vol. 1, ed. Max Lütolf (Bern: Francke Verlag, 1978), item 214. ⁶⁹ . Dreves et al., eds., Analecta Hymnica, vol. 51, item 84, pointed out by Dag Norberg, An Introduction to the Study of Medieval Latin Versification, trans. Grant C. Roti and Jacqueline de la Chapelle Kubly, ed. and intro. Jan Ziolkowski (Washington: The Catholic University of America Press, 2004), 32. Norberg suggests assonance rhyme in Medieval Latin poetry may result from the influence of Sedulius, and see also the comments by Andy Orchard in The Poetic Art of Aldhelm (Cambridge: Cambridge University Press, 1994), 40. There are, however, other possibilities. Gasparov believes that rhyme developed in European poetry when exiled Irish authors transferred their Celtic rhyming habits to Latin verse, while Lanz argues that, since rhyme is simply another form of parallelism, spontaneous generation cannot be excluded. See M. L. Gasparov, A History of European Versification, trans. G. S. Smith and Marina Tarlinskaya, ed. G. S. Smith and Leofranc HolfordStevens (Oxford: Clarendon Press, 1996), 98; and Henry Lanz, The Physical Basis of Rime (Stanford: Stanford University Press, 1931), 125 and 127–8. ⁷⁰ See Matonis, “English and Welsh,” 6–7 and 13–21. Though some of the Harley lyrics were clearly imports, this particular lyric appears to have been composed in the Welsh border area; see Frances McSparran, “The Language of the English Poems,” in Susanna Fein, ed., Studies in the Harley Manuscript: The Scribes, Contents and Social Contexts of British Library MS Harley 2253 (Kalamazoo: TEAMS, 2000), 418–20. For the provenance of the manuscript itself, see Carter Revard, “Scribe and Provenance,” in Fein, Studies, 21–109. ⁷¹ Given the evidence for a Welsh presence at the English court (Matonis, “English Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 22 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 649 gues that “The use of rhyme within Church Latin from the fourth to the thirteenth centuries accounts for the appearance of rhyme not only in English verse, but in other vernacular literatures of Europe,”⁷² while Macrae- Gibson notes that such a connection can hardly be doubted in the case of Cynewulf, where end rhymes seem likely to reflect Cynewulf’s acquaintance with the Latin hymnody of his time.⁷³ Rhymed Latin verse was cultivated in England from the seventh century onwards, and rhymed verse in Latin was clearly familiar enough for the seventh- century archbishop of Canterbury, Theodore of Tarsus, to use it to request prayers from a bishop.⁷⁴ Moreover, as far as the four by four pattern in particular is concerned, it is worth remembering that, as Lapidge observes (and as the above examples illustrate), “The characteristic form of a Latin hymn was the iambic dimeter: a very compact form in which each line consisted of four feet, of which the second and fourth were necessarily iambic. . . .Four iambic lines thus constituted were combined to make up a stanza.”⁷⁵ This is basically verse in the four by four formation, and, while the rhyme scheme varied, the abab pattern was certainly not unusual. See, for instance, the following from volume 1 of the Analecta Hymnica: Divo falgrans numine Maria dulcoris Congaudet in culmine Angelicis choris. (item 74 and see also items 80, 89, 117, 130, 204, 214 etc.)⁷⁶ In the case of one particular form of English verse, the carol, there is, moreover, very clear evidence that the use of the abab stanza did result and Welsh,” 2 and 4), it seems possible that Welsh bardic influence may sometimes have spread beyond the border area, but there is little evidence of this in the abab poems, which lack examples of the typically Welsh forms of alliteration, cymeriad and cynghanedd identified by Matonis in the Harley Lyrics (Matonis, “English and Welsh,” 13–20). ⁷² Michael McKie, “The Origins and Early Development of Rhyme in English Verse,” Modern Language Review 92 (1997): 817–31. ⁷³ O. D. Macrae-Gibson, ed., The Old English Riming Poem (Cambridge: D. S. Brewer, 1983), 23. See also the discussion by Milfull, The Durham Hymnal, 1 and passim. ⁷⁴ Viz.: “Te nunc sancte speculator / Verbi dei digne dator / Hæddi pie presul precor / Pontificum ditum decor / Pro me tuo peregrino / Preces funde Theodoro.” Cited by Macrae- Gibson, Riming Poem, 24, and see also A. W. Haddan and W. Stubbs, Councils and Ecclesiastical Documents Relating to Great Britain and Ireland, vol. 3 (Oxford: Oxford University Press, 1871), 203. ⁷⁵ Michael Lapidge, Bede the Poet, Jarrow Lecture 1993 (Jarrow: 1994), and see also the comments by Richard Leighton Greene, The Early English Carols (Oxford: Clarendon Press, 1935), lxxi. [AU; NO PRESS FOR THE LAPIDGE?] ⁷⁶ Dreves et al., eds., Analecta Hymnica . Sheet 22 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 23 of 257 650 Romances in the abab Tradition from Latin influence, since Greene provides examples of such carols in which vernacular a-lines combine with b-lines which are actually taken from Latin hymns.⁷⁷ The abab stanza thus grew out of a tradition of sung verse, verse which would be likely, as we have seen, to be stricter as regards vowel identity than consonant identity, and while rhymes in some other types of English poetry, influenced perhaps by the gradual move in the direction of the written and the read rather than the oral, would gradually lose this flexibility, the four by four stanza, which has always retained its association with the spoken and the sung, appears not to have done so. Not surprisingly, this seems to have been true of other verse forms which retained their links with the oral. Middle English carols with a variety of stanza forms can have rhymes which are imperfect in respect of their consonants,⁷⁸ while Masa Ikegami has found examples of assonance in the play Mankind, the only instances of such rhymes in his corpus,⁷⁹ including a large number of /n/:/m/ rhymes, as well as rhymes between plosives, fricatives and alveolars (including /p/:/k/ rhymes of the type we considered earlier). Rhyme of this type is also found in Havelok,⁸⁰ and though this poem was clearly a written rather than an oral composition (see ll.2999–3000), the acceptance of imperfect consonantal rhyme, together with comparative strictness as far as rhyming vowels are concerned, may well suggest something about the expected reception of the poem (i.e., that it would be read aloud to an audience).⁸¹ More controversially, such rhymes might also provide some support for the view that the author may have been drawing on oral versions ⁷⁷ Greene, The Early English Carols, lxxii. ⁷⁸ See, e.g., Greene, The Early English Carols, items 2.stanza 2, 12.1, 31.4, 39.1, 40.3, 50.1, 60.5, 78.9 (compare previous stanzas), 85.3, 94.1, etc. ⁷⁹ Ikegami, Rhyme and Pronunciation, 179–80. ⁸⁰ See the edition by G. V. Smithers (Oxford: Clarendon Press, 1987), lxxiii–lxxiv. ⁸¹ The view that the Havelok MS was a minstrel’s copy has long since been disproven (see John C. Hirsch, “‘Havelok’ 2933: A Problem in Medieval Literary History,” Neuphilologische Mitteilungen 78 [1977], 339–47): but this does not mean that the poem was not delivered orally. On oral delivery of romances, see Albert C. Baugh, “The Middle English Romance: Some Questions of Creation, Presentation, and Preservation,” Speculum 42 (1967): 1–31 (in which he argues convincingly that although romances were often written compositions, they were nevertheless designed for a listening audience), and for similar views on the oral delivery of romance, see Karl Reichl, “Orality and Performance,” in A Companion to Medieval Popular Romance, ed. Raluca L. Radulescu and Cory James Rushton (Cambridge: D. S. Brewer, 2009), 132–149, esp.135–8; and Ad Putter, “Middle English Romances and the Oral Tradition,” in Medieval Oral Literature, ed. Karl Reichl, (New York: De Gruyter, 2012), 335–51. Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 24 of 257 Judith A. Jefferson, Donka Minkova, and Ad Putter 651 of the story.⁸² The link with the oral means that it is clearly unhelpful to regard this tradition of “false rhyme” as inferior, although such an attitude has often been adopted; Jean Louis Aroui, for instance, has observed that, “It is . . . probable that, for many languages, the strict homophony of rhyme is a kind of aesthetic ideal, which is just imperfectly triggered when the formal ability of the poet (or the performer) has not been stimulated by a learned acquisition.”⁸³ Given the widespread use of rhymes of the type we have been considering, strict rhyme cannot be said to have constituted the aesthetic norm, nor should departures from it be seen as incompetence. The use of feature rhyme, for instance, requires considerable, if probably unconscious, knowledge of the way in which phonemes are related to one another, and Middle English poets in the abab tradition made good use of this knowledge to create rhymes that would be acceptable to their audience.⁸⁴ Judith A. Jefferson, University of Bristol Donka Minkova, University of California, Los Angeles Ad Putter, University of Bristol ⁸² For an argument strongly in favour of an English oral rather than a French written source for the poem, see Nancy Mason Bradbury, “The Traditional Origins of Havelok the Dane,” Studies in Philology 90 (1993): 115–42. It is not necessary, however, to agree wholeheartedly with Bradbury in order to accept that the oral tradition may have had at least some influence. See, e.g., the comment by Pearsall that “the English romance [i.e. Havelok] is a free reworking. . . . . of some variant of the French poem. . . . .with contamination from local Lincolnshire tradition,” in Derek Pearsall, “The Development of Middle English Romance,” in Studies in Medieval English Romances: Some New Approaches, ed. Derek Brewer (Cambridge: D. S. Brewer, 1998), 18. ⁸³ Aroui, “Proposals for a Metrical Typology,” in Towards a Typology of Poetic Forms, ed. Jean Louis Aroui and Andy Arleo (Amsterdam: John Benjamins, 2009), 26. ⁸⁴ This article is the result of research into the verse forms of ME romances for which we have received funding from the Arts and Humanities Research Council, whose support we gratefully acknowledge. Sheet 24 of 257 Tseng Proof • 2014.06.18 09:58 9555 Studies in Philology • 111:4 • Sheet 25 of 257