Linguistic variables are units within a language which vary in a socially significant way and are thus “most subject to social or stylistic variation” (Crystal 2003:488). Each linguistic variable will have a set of alternative realisations, known as variants, which are semantically equal. Even though not all linguistic units can be classified as variants, those which are do not occur randomly. Instead, they can be said to vary “systematically in relation to other variables, such as social class, age and gender” (Coates 2004:48) and are involved in co-variation. Therefore, if a speaker chooses to say [§u:n] or [fu:] rather than [tju:n] or [fju:], certain social and/or regional information relevant to that person is revealed.
The variable (ju) – A Historical Perspective
In the seventeenth century, the diphthong [Iu#] (where the diacritic “ #” indicates extra shortness) was present in English and served to distinguish lexical items such as threw and brewed from through and brood which contained the monophthong /u:/. These words containing the monophthong /u:/ had derived earlier, as a result of the Great Vowel Shift, from Middle English /o:/ around the fifteenth and sixteenth centuries. However, items such as dupe, huge and music “contained a variety of Middle English vowels (/Iu, Eu, y:/ etc.), all of which had merged by the seventeenth century into a falling diphthong of the [Iu] type” (Wells 1982a:206). The [Iu] diphthong at this stage was akin to current primary diphthongs such as /aU/ where the syllable peak is the initial element and is followed by an off-glide (Roca and Johnson 1999:276). However, by the end of the seventeenth century, the peak had been transferred to the second element, while the initial element [I] was raised to [i] leaving [i#u]. With the syllable peak apparent on the second element, Roca and Johnson note that “the assignment of the [i] to the onset follows automatically as a matter of formal necessity” (1999:277). This is illustrated by the use of /j/ (or /y/ in North America) in the transcriptional representation.
Subsequently, Early Yod Dropping (EYD) saw the disappearance of the /j/ in certain environments acting as a prelude to Later Yod Dropping (LYD) which eliminated /j/ from historical /ju/ positions, such as before /t, d, s, T, n/. Although EYD did not affect some accents (such as Welsh and American English), the phonological environments most effected were /ju/ following palatals and palato-alveolars, /r/ and consonants plus /l/ (Wells 1982a:2006-207). Most notably, accents associated with East Anglia are recognised as having Generalised Yod-Dropping (GYD) which covers most or, in some cases, all post-consonantal environments.
(3.1) From Wells Accents of English Volume 1 – An Introduction; 1982a:207
a. EYD b. GYD
j ¼ / r __ j ¼ / C__
/ju/ - In Unstressed Syllables
The discussion so far has referred to the status of /ju/ when located in stressed or strong syllables (such as tu!ne) as well as syllables where there is no possibility of vowel reduction due to secondary or tertiary stress (such as a!ttitu~de).
However, with respect to unstressed syllables containing the onsets /t/ and /d/, elimination of the palatal /j/ is less widespread with the advent of a merger, or coalescence between /tj/ and /dj/ resulting in the affricates /§/ and /½/, respectively. The transition from /tj/ and /dj/ to these affricates is argued by Bauer (1994) to be counted as yod-retention as opposed to yod-deletion. This is due to the retention of the palatality associated with /j/ in the merged forms, as well as the view that affrication is generally accepted to be a strengthening process or fortition, and that deletion is the absolute form of weakening or lenition. Thus, the implication is that “yod-dropping and coalescence have opposite effects, not the same effect” (Bauer 1994:109).
Yod-dropping can therefore be considered as phonetically abrupt, in keeping with the Neogrammarian principles of sound change, which are characteristically regular and abrupt. However, Bauer (1994:165) observes that a palatalised [tÆ], for example, would phonetically constitute an intermediate form between items such as [tju:n] and [tu:n], although it is likely to be misperceived as belonging to either [tj] or [tu]. Therefore, with respect to this study, the variable (ju) is represented through the variants [ju], [u] and, in the event of a preceding /t/ or /d/, [§] and [½].
/ju/ - A Phonological Perspective
The sequence /ju/ consists of a glide /j/ plus a close back vowel /u/. The glide element /j/ can be described as a vocoid and is labelled by Gimson (2001:211) as an unrounded palatal approximant which, with respect to its articulation, uses the tongue to assume the position of a close-mid to close-front vowel before immediately moving to the position of the following sound. The significant distinction between high vowels and glides is how they are represented in their syllabification. Thus, with respect to the palatal approximant, syllable nuclei will represent the high vowel [i], while its manifestation in the margins will be [j]. Indeed, Baertsch (2004) suggests that [i] is a coercible margin and that both high vowel and corresponding glide are, underlyingly, featurally identical, while the surface distinction stems from an underlying length distinction.
The distribution of /j/ as a margin glide is restricted to only the onset position of a syllable, since it becomes a diphthong when preceded by another vocoid. By way of illustration, a high vocoid followed by another vocoid would produce, for example,
[iEs] → [jEs] ‘yes’, whereas a high vocoid preceded by another vocoid is represented in such a way as, for example, [tai] → [taj] → [taI] ‘tie’.
When the glide /j/ occupies the onset, it can be observed to pattern with the consonants rather than vowels. Evidence of its phonemic status occupying this position is provided through minimal pairs such as yacht~cot and yes~guess. Thus, when /j/ maintains a simple onset, it can be followed by any vowel (hence yacht, yet, yak etc) due to the claim that “co-occurrence restrictions do not hold between onset and rime constituents” (Borowsky 1986:277). However, if the glide is the right-most element in a branching onset (disregarding /s/-clusters), the nuclear vowel must be [u], unless vowel reduction manifests [«] or [U]. Therefore, ‘true’ occurs as [t¨u:] but not *[t¨ju:]. This, in addition to /ju/’s inability to follow voiced fricatives, such as *[vju:], (a trait shared by the other approximants of English /l r w/), implies that /ju/ is a diphthong rather than an onset plus nucleus.
nb - This discussion regarding the collected data is continued in section seven below. Phonological Analysis