Dating the historical SVLR
Lass (1974: 320) sees the historical SVLR as bipartite, as shown in (1).

(1) (a) All long vowels shortened everywhere except before /r v z Ʒ ð #/

     (b) The nonhigh short vowels /ε a ɔ/ lengthened in the same environments.

The effect of SVLR is clear: before its operation, Scots, like other Middle English dialects, contrasted long and short vowels, whereas afterwards, Scots had innovated a system in which length is non distinctive. Pullum (1974) argues that this reanalysis of the underlying Scots vowel system results directly from the introduction of SVLR, observing that: `an immediate or even simultaneous consequence of the addition of a rule like Lass' formulation of Aitken's Law (a) to a grammar would be a restructuring by rule inversion: from underlying vowels shortened in all contexts except before /r v ð z Ʒ #/, the language would shift to having underlying short vowels lengthened before /r v ð z Ʒ #/'.

That is, the transition from historical to SVLR involves a classic case of rule inversion (Vennemann 1972), which is, of great relevance to the application of Lexical Phonology to sound change. (2) shows the input and output systems for SVLR, to illustrate Pullum's proposed change.
(2) 

The output system in (4.21) also provides further support for the laxing and lowering of short vowels which was required above to account adequately for MEOSL. It is clear that the Modern Scots and SSE vowel system must include /ε a ɔ/, since each occurs in a fairly large set of lexical items (men, bed, slept for /ε/; cot, caught, pot, law for /ɔ/; and back, trap, car for /a/). In order to derive such a system historically, the short vowel system prior to the operation of SVLR cannot have been that of Old English, i.e. /i e a o u/, since the requisite length adjustments of SVLR would then have produced mergers of /i/ with /i:/, /u/ with /u:/, /e/ with /e:/, /o/ with /o:/ and /a/ with /a:/, and no source would be available for /ε/, while /ɔ/ would remain extremely marginal. We must rather assume a short lax vowel system /ɪ ε Λ ɒ a/, with /Λ/ having lowered from earlier /ʊ/ after GVS. Lass assumes that /a/ then merged with earlier /a:/ and /¡/ with earlier /ɔ:/, while /ε/ lengthened in the appropriate SVLR long environments, fitting into the same system as the originally long vowels as a new underlyingly short vowel with contextually long realizations. In fact, the status of /ε/ vis-à-vis SVLR is unclear; we shall also consider reasons for the exceptionality of /ɪ Λ/, which do not lengthen either synchronically or historically. As (4.21) shows, SVLR disrupted the correlation of length with [+ tense], and short with [-tense], which had held in English since around the time of MEOSL.

A fairly wide spread of dates for SVLR has been proposed: Lass (1976: 54) opts for the seventeenth century; McClure designates SVLR as ‘a sixteenth-century sound change in Scots' (1977: 10); and Aitken half commits himself to an earlier introduction ‘? in the fifteenth century' (1981: 137). Although it is not possible to be conclusive, evidence of several types suggests that SVLR was under way by the mid to late sixteenth century.

Johnston (1980: 380) opts for the period 1600 - 1640, stipulating that SVLR must follow GVS and precede lowering of /ʊ/; but neither of these claims is well supported. First, Johnston argues that /ʊ/ lowered to /Λ/ before SVLR because he assumes that the descendants of Older Scots /i u/, which failed to lengthen by SVLR, are exempt on account of their height. However, the originally long high vowels /i: u:/ were affected by SVLR; and the date (from Dobson 1962) which Johnston accepts for lowering may also be rather late: the process may well have been sixteenth rather than seventeenth century in the North, since it is generally assumed to have operated immediately after the GVS, the last stage of which Johnston himself dates to c.1500 - 1550 in the North, although over a century later in the South. As far as interaction of SVLR with GVS is concerned, Harris also asserts that `the shortening of historically long vowels ... post-dates the early stages of the Great Vowel Shift, since these vowels all appear in their shifted shapes' (1985: 23). Thus, divine has short [Λi] in Modern Scots and SSE, shifted from earlier /i:/; similarly, meat has [i] from pre-GVS /e:/, and coal, [o] from /ɔ:/: if SVLR had preceded GVS, these vowels, in SVLR short contexts, would have been short and therefore ineligible for shifting. There is, however, persuasive evidence that SVLR and at least part of the GVS were contemporaneous. Recall that /i:/ is generally taken as having shifted initially to [εi], [Λi] or [əi] by GVS, on the grounds that a direct shift to [ai] would predict an unwarranted merger with /ai/, which raised only later. This theoretical motivation for the intermediate shift is strongly supported by the orthoepical evidence, in which an open front first element is not attested until the 1740s (Lass 1989, in press). In Modern Scots and SSE, Older Scots /i:/ in SVLR long environments gives the long diphthong [a:i], whereas in short contexts we still find [Λi]; McClure (1995: 51) interprets this as indicating that `the diphthongization of /i:/ by the Great Vowel Shift had begun before the operation of Aitken's Law, but was arrested in words where the Law resulted in shortening and carried to completion only in those where the vowel remained long.'

Some further evidence for a relatively early dating of SVLR comes from Harris (1985: 23). SVLR operates in Ulster Scots, at least for some vowels, and Harris argues that, since most Scottish settlers of Ulster migrated from the peripheral dialect areas of southwest Scotland during the Plantation of Ulster from 1601 onwards, `the Aitken's Law changes must presumably have begun their diffusion outwards from the core dialects of central Scotland well before the seventeenth century if they were to be sufficiently advanced in southwest Scots before the Plantation of Ulster.'

Approaching the issue from the other end, it is clear that SVLR must have been well established by the eighteenth century. First, McClure (1995: 51) cites Shetland dialects where /d/ and /ð/ had merged by this time; however, vowels are long before [d̪] /ð/, historically a lengthening environment, and short before [d̪] /d/, as in [mi:d̪] meed `landmark' versus [nid̪] need. Primary sources from the eighteenth century, although rarely explicit, also testify to the general Scots neutralization of length, in frequent comments to the effect that Scots cannot reproduce English vowel length distinctions. Thus, Buchanan (1770: 44) rather wearily tells us that `I shall adduce but a few examples, out of a multitude, to shew how North-Britons destroy just quantity, by expressing the long sound for the short, and the short for the long'.

Similarly, Drummond (1767: 21) asserts that `The sound of every vowel may be made long or short ... But to ascertain the time of pronouncing them is the greatest difficulty to the Scots, in the English Tongue.'

Jones (1997b: 294) extracts from various eighteenth century sources a series of characteristic Scots pronunciations, involving the alternations [i] ~ [ε], [e] ~ [a] and [o] ~ [ɒ], which `appear to represent height contrasts, not unlike those characteristic of the Great Vowel Shift process itself, but in contexts where the affected vowels do not appear to meet the expected (and necessary) extended length criterion so intimately associated with it'.

This observation may provide further evidence of the interaction of SVLR with GVS: if the two overlapped chronologically, we might expect some extension of GVS to etymologically short vowels, at least in SVLR long environments. For instance, Elphinston (1787: 14) notes that Scots has mak, tak, brak, mappel, apel, craddel, sadel, as opposed to English make, take, break, mapel, appel, cradle, saddel,and gairden, yaird, dazel, staig, compared with English garden, yard, dazzel, stag. In the first set of forms, there seems to be raising in English but not in Scots, while the second set shows the opposite pattern; and in both cases, there is a very strong, though not exceptionless, correlation with SVLR short and long environments respectively.

Occasionally, an eighteenth century source is more explicit about vowel length in Scots. Two examples worthy of mention are Sylvester Douglas and Alexander Scot (see Jones 1991, 1993, 1995). We know rather little about Alexander Scot (or Aulaxaunder Scoat, as he signs himself), whose letter The Contrast, dated 1779, is the subject of Jones (1993). The Contrast is directed to a noble family, and describes the impressions of a Scot on returning to Scotland; in particular, Scot comments on changes, including linguistic ones, which he observes.

The important aspect of The Contrast for present purposes is Scot's use of five different orthographic representations for English [ai] (3).
(3) 

Leaving aside the little-populated classes 3 and 5, Scot's system captures the environments of SVLR, with class 4 for short contexts, involving following voiceless consonants, voiced stops, /l/ and /n/, and classes 1 and 2 covering the long contexts. As we shall see below, experimental evidence demonstrates that final position produces longer vowels than following voiced fricatives and /r/.

The second source of explicit eighteenth century comment on Scots vowel length is Sylvester Douglas (Jones 1991), whose Treatise on the Provincial Dialect of Scotland (containing `a table of words improperly pronounced by the Scotch, showing their true English pronounciation' (Jones 1991: 158)) was written in 1799. Douglas was born in 1744 near Aberdeen, and trained first in medicine, then in the law, becoming a barrister in 1776 and KC in 1793. Between 1795 and 1806 he was an MP for various constituencies, and in 1800, was elevated to the Irish peerage as Baron Glenbervie. He did not succeed in his ambition to rise to the English peerage, or to take up a cabinet post, although he did marry a daughter of Lord North (albeit, according to a commentator of the time, not a terribly attractive daughter). Douglas's Treatise contains various comments on vowel length in Scots. For instance (Jones 1991: 124), he mentions a Noun - Verb alternation of voicing which still carries with it an SVLR distinction of length for Modern Scots: `In thief the ie has the first sound of the e shortened. Add the e at the end and as in thieve (where indeed the consonant is also altered) and the ie retains the same sound but protracted.'

Under EASE (Jones 1991: 190), Douglas gives more detail on the same vowel: `The ea as in appear the s soft as in please ... pronounce in the same manner, appease, disease, please, tease, lease verb synonymous with glean. In the following words the ea has the same sound but shortened and the s is hard: cease, decease, surcease, lease verb and noun ... release ...'

Although Douglas's comments are enlightening, they are incomplete; the Treatise is not wholly systematic, and the alphabetically arranged pronouncing dictionary within it is extremely selective, so that the items which might be most informative often do not appear. In interpreting eighteenth century Scottish sources, we must also remember that the authors generally had at least one eye on the favored English system. Jones (1991: 4) has `no strong impression that Douglas is advocating the total abandonment of all Scottish vernacular features in favor of an undiluted London upper class norm'; but Douglas nonetheless subtitles his Treatise `an attempt to assist persons of that country in discovering and correcting the defects of their pronunciation and expression'.

مواضيع ذات صلة

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

Justifying the S in SVLR

The Scottish Vowel Length Rule

The Great Vowel Shift

Middle English Open Syllable Lengthening

Internal history

Vowels

Consonants