Middle English Open Syllable Lengthening
Middle English Open Syllable Lengthening (MEOSL) remains a controversial process, and one discussed in detail by Dobson (1962), Malsch and Fulcher (1975), Lieber (1979), Minkova (1982, 1985) and Ritt (1994); see also Hogg (1996), who argues that a regular but allophonic low-vowel Open Syllable Lengthening in Old English acted as `the first harbinger' (1996: 70) of MEOSL. It is beyond contention that all short vowels lengthened in the North, and [-high] short vowels in the South, in certain environments around the early thirteenth century; some examples are given in (1).

(1) 

However, various aspects of MEOSL are still disputed, and Minkova (1982: 29) singles out three of the more problematic areas:
(1) The problem of the qualitative difference between the original short vowels and their lengthened reflexes.
(2) The behavior of the high vowels /i/ and /u/ with respect to the change.
(3) The existence of a large number of exceptions to MEOSL.

Minkova herself deals with the third problem. She notes that, given the `traditional' environment of /Ð C2 1VC0# (which allows for medial /sp st sk/), a large number of exceptions to MEOSL become apparent. Authors of Middle English handbooks, like Jordan (1925) and Luick (1921), and also, more recently, Dobson (1962), have attempted to account for these exceptions by noting that many contain a liquid or nasal in the second syllable, or by grouping together items like bodig, popig, penig, hefig and postulating either secondary stress on-ig or long final /i:/. Sadly, these attempts at principled explanation are either non-explanatory (as in the first example above) or lack evidence (as in the second).

Minkova adopts a different approach, based on a complete list of words which are known to have been present in English at the time of MEOSL, meet the structural description of the process, and have survived to Present-Day English. She includes only items with original non-high vowels, since /i u/ lengthened inconsistently, and considers both native and Anglo-Norman material. Minkova splits the items on her word-list into two sets, one containing items which are still disyllabic in Modern English and the other composed of items which are now monosyllabic due to final schwa-loss, and calculates the percentage of the words in each set which have undergone MEOSL. She finds that only 16 per cent of the synchronically disyllabic words exhibit lengthening, while MEOSL has operated without exception in words which have also undergone schwa-loss. Minkova concludes that `it is surprising that so much energy has been expended in trying to account for ``exceptions'' which make up over 80% of the entire material' (1982: 42), and recasts the environment for MEOSL (2).
(2) 

This reformulation indicates a definite link between MEOSL and schwa-loss, but does not determine their relative chronology: indeed, Minkova (1982: 46) argues that `simultaneity is the only positive assumption we can make'. Jespersen (1922) also suggested that schwa-loss began in the North due to Norse influence, which was strongest in this area, and a concomitant loss of inflection. No case of this sort has been made for MEOSL, but the intimate connection of MEOSL and schwa-loss assumed by Minkova predicts that MEOSL too should have started in the North, since schwa-loss, a prerequisite for the lengthening, is first evidenced in this area; both processes subsequently spread south over a century or so.

Minkova (1985) develops her analysis of MEOSL in terms of foot structure, arguing that lengthening preserves perceptual isochrony of feet when these become defective as a result of schwa loss. This accounts for the predominant application of MEOSL in original disyllables, and the greater likelihood of lengthening in rhymes with non-branching peaks or codas. Ritt (1994), in turn, extends Minkova's work: using a parallel but larger corpus of Middle English items, he establishes that 94 per cent of items with unstable final syllables (that is, those subject to schwa loss) undergo lengthening. Ritt, however, includes MEOSL in a general schema of Middle English Quantity Adjustment, producing a composite statement of the probability of lengthening in any case based on a set of scalar features. For instance, Ritt shows that lengthening is more likely for low and back vowels; more likely before a single consonant than a cluster; and less likely in a syllable weighing two moras or less.

Neither of these environmental refinements, however, addresses the problem of `the qualitative difference between the original short vowels and their lengthened reflexes' (Minkova 1982: 29). Two main sources of evidence indicate that MEOSL produced a qualitative as well as a quantitative change for non-low vowels. First, in twelfth and thirteenth century manuscripts, orthographic alternations are found between un inflected and inflected forms of words (see (3)).

(3) 

These spellings do not suggest simple lengthening; the fact that alternates with and with indicates that the Old English short vowels have both lengthened and lowered. Such evidence is available only for the high vowels, since the long high-mid vowels /e: o:/ and the long low-mid vowels /ε: ɔ:/ were not orthographically distinguished in Middle English, /e: ε:/ being written and /o: ɔ:/, . A second category of rhyme evidence is more relevant to the mid vowels. If MEOSL involved only a quantity change, one would expect the lengthened reflexes of Old English /i u e o/ to rhyme with Middle English /i: u: e: o:/ respectively. Attested rhymes (4) do not show this pattern. Instead, they again indicate that MEOSL involved a quality change, whereby short high vowels in open syllables merged with long high-mid vowels, short high-mid vowels in the MEOSL environment merged with long low-mid vowels, and only /a/ merely lengthened.

(4) 

Some additional evidence for this proposed quality change comes from the Great Vowel Shift. If our usual assumptions about this sound change are correct, the relevant non-low vowels must have lowered before shifting (Malsch and Fulcher 1975). Week, for instance, surfaces in Modern English with [i:]; this is consistent with its having had an /e:/ vowel in Middle English, but not /i:/, which would have produced modern /aɪ/. Similarly, bear, with Modern English [e:]/[εə], must have had /ε:/ at the time of operation of the Great Vowel Shift; if Old English /e/ had simply lengthened to /e:/ in Middle English, one would expect Modern English *[bi:r] or *[bɪə] `bear'. The effects of MEOSL are schematised in (5).

(5) 

As Ritt (1994: 76) observes, `one cannot avoid dealing with vowel quality as well, if a change in quantity causes, goes hand in hand with, or, generally speaking, is closely related to a quality change'. But there seems to be nothing inherent in a lengthening process like MEOSL that should lead to concomitant lowering; Pre-Cluster or Homorganic Lengthening, a tenth century vowel lengthening change, appears to have had no effect on quality (Ritt 1994: 82). One solution might be to identify some process affecting the set of short vowels, from which both Pre-Cluster Lengthening and MEOSL took their inputs, between the tenth and twelfth centuries. The obvious assumption would be that non-low short vowels lowered during this period; Dobson (1962) actually proposes that /i u/ had become high-mid and /e o/ low-mid before MEOSL, and quotes arguments by Trnka and Vachek to the effect that the resultant isolation of /i: u:/, which were left with no short congeners, caused them to move out of the monophthongal system altogether by diphthongizing during the Great Vowel Shift. However, there is no direct evidence for this lowering.

An alternative explanation might be that the feature [± tense] became relevant in the English vowel system at around the time of MEOSL. Ritt (1994) proposes that Middle English long vowels were made tense; lengthened vowels would then merge with the long, tense vowel one degree of height lower. As Ritt (1994: 79) summarizes: `Although I lack positive evidence in much the same way as all other linguists who have dealt with the apparent lowering that accompanied Middle English vowel lengthening, it seems to me that this is by far the most economic explanation of the notorious mergers'.

There are two difficulties with such an analysis. The first involves the use and status of the feature [tense] itself, and particularly the question of its phonetic correlates. For the moment, let us simply note that lax vowels are characteristically produced with a lesser degree of constriction than their tense counterparts. X-ray evidence from Wood (1975: 110) indicates that the tongue height of a long, lax vowel will tend to be closer to that of a long, tense vowel of one degree less in height than to that of a long, tense vowel of the same height. Indeed, lax [ɪ:] was often produced with a lower tongue height than tense [e:] in Wood's experiments. Consequently, if we have short lax vowels, and long tense ones, then a lengthening of the lax ones might well involve automatic lowering. It seems worthwhile taking [tense] on trust briefly; however, a detailed justification follows, since variability in the length ± tenseness correlation will be central to the analysis of SVLR.

Secondly, we require evidence for some quality-based realignment of the Middle English vowel system, whatever feature(s) we choose to label this. Essentially, factors in the suprasegmental organization of the language may account for the innovation of [tense] at just this period. During Middle English, the English stress system was undergoing a radical change, with the introduction of the phonologically determined Romance Stress Rule via French loans, alongside the earlier Germanic Stress Rule, which was morphologically determined and assigned main stress to the first syllable of each stem. Although syllable weight as a phonological variable appears to have existed in Old English, for instance as a factor determining the assignment of secondary stress, the introduction of the Romance Stress Rule initiated a more pervasive correlation between syllable weight and stress; the rule scans words right to-left, and preferentially stresses heavy syllables (final in verbs, penultimate in nouns). If the first relevant syllable is not heavy, the stress is placed on the previous syllable, regardless of weight.

Hyman (1977: 47-9) notes that languages with a stress assignment system making reference to syllable weight always have a vowel length contrast (although length contrasts per se are not confined to languages with phonologically determined stress rules, as Old English and Polish demonstrate). And in languages where syllable weight is a phonological variable and there is a length contrast, there is almost always a quality distinction between long and short vowels of the `same' height. When English, which already had a vowel-length contrast, borrowed the Romance Stress Rule, which refers to syllable weight, it might therefore be expected to acquire a tenseness, and thus a quality distinction between long and short vowels. Anderson (1984) further argues that languages will tend to implement a redundancy rule which correlates underlying length, or nuclear complexity, with phonological tenseness.

The question remains whether the innovation of [tense] preceded or followed MEOSL. There is some dialect and rhyme evidence for the persistence, at least initially, of two sets of long vowels, which one might see as lax and tense. For instance, in Yorkshire, Derbyshire, Lancashire, Staffordshire and other northern and north Midland areas of England, [ε:] and [ɔ:] from MEOSL have become Modern English [eɪ] and [oi], while older /ε:/ (<OE /æ:/) and /ɔ:/ (<OE /ɑ:/ and French loans) have developed to [iə] and [uə]. These facts suggest that `the reflexes of short vowels in open syllables and of original OE long vowels were still kept distinct in the Middle English dialect from which the modern north Midland and northern dialects derive' (Lieber 1979: 16). In terms of rhyme evidence, Dobson (1962) quotes a stanza from Troilus and Cryseyde, with rhyme scheme ABABBCC, which exhibits the rhymes shown in (6).
(6) 

According to Dobson, it is inconceivable, given Chaucer's rhyming practice, that he would have rhymed all five consecutive lines as A. This suggests that the A-line vowels and the B-line vowels must have been distinct at this time, perhaps as [ɔ:l] versus [ɔ:t]. One might then speculate that the correlation of short and lax was implemented earlier than that of long and tense, though little hangs on establishing a chronology. The main point here is that assuming a qualitative bisection of the Middle English vowel system, in addition to pre-existing distinctions of length, is helpful in accounting for the lowering associated with MEOSL of non low vowels. I ascribe this qualitative difference to tenseness, and will justify this decision below in connection with SVLR.

مواضيع ذات صلة

Dialect differentiation in Lexical Phonology: the unwelcome effects of underspecification Introduction

The life-cycle of the Scottish Vowel Length Rule

From sound change to phonological rule Standard Generative Phonology and Lexical Phonology

The ordering of SVLR and LLL in a Lexical Phonology

The environment for SVLR

Evidence for the interaction of SVLR and LLL

Low-Level Lengthening

The Scottish Vowel Length Rule and Low-Level Lengthening Experimental evidence

The input to the Scottish Vowel Length Rule

The Scottish Vowel Length Rule in Present-Day Scots and Scottish Standard English

SVLR and diphthongs

Dating the historical SVLR