Phonetic parallels between the close-mid vowels of Tyneside English

Abstract

The distribution of variants of the FACE and
GOAT vowels in Tyneside English (TE) is assessed
with reference to the age, sex, and social class of 32
adult TE speakers. The effects of phonological context
and speaking style are also examined. Patterns in the data
are suggestive of dialect leveling, whereby localized speech
variants become recessive and pronunciations typical of
a wider geographical area are adopted. Within this broad
pattern, however, there is evidence of parallelism between
the vowels in terms of the relative proportions of their
variants across speaker groups. It is suggested that pressure
to maintain the symmetrical structure of the underlying
phonological system is guiding this process. Labov's
(1991, 1994) principles of chain shift are discussed in
this connection. However, it is argued that the patterns
in the data are more plausibly explained by considering
the social significance of each variant instead of making
reference to variants as socially neutral expressions of
abstract phonological categories.

Full text

Phonetic parallels between the close-mid vowels
of Tyneside English: Are they internally
or externally motivated?
Dominic J. L. Watt
University of York
ABSTRACT
The distribution of variants of the face and goat vowels in Tyneside English (TE)
is assessed with reference to the age, sex, and social class of 32 adult TE speakers.
The effects of phonological context and speaking style are also examined. Patterns
in the data are suggestive of dialect leveling, whereby localized speech variants
become recessive and pronunciations typical of a wider geographical area are
adopted. Within this broad pattern, however, there is evidence of parallelism be-
tweenthe vowelsin termsof therelative proportions oftheir variantsacross speaker
groups. It is suggested that pressure to maintain the symmetrical structure of the
underlying phonological system is guiding this process. Labov’s (1991, 1994) prin-
ciples of chain shift are discussed in this connection. However, it is argued that the
patterns in the data are more plausibly explained by considering the social signifi-
cance of each variant instead of making reference to variants as socially neutral
expressions of abstract phonological categories.
Certain tensions underlie current theories of the dynamics of vowel systems,
especially those developed to model chain-shift processes.
1
Chain shift, indeed,
is predicated upon the interaction of mutually antagonistic forces (preservation
of contrast vs. parsimony of oppositions), with the contribution of the speaker
being to act as a sort of conduit for the transmission of change once it has been set
in train. Strong internalist accounts of vowel system dynamics have tended to
marginalize the role of speakers, preferring instead to portray shifts as long-term,
teleologicalprocessesextending sometimesover severalgenerations. Inthisview,
they are processes over which speakers have little or no control. The role of the
speech community here could be said at best to facilitate the spread of a change
and to circumscribe the limits of that spread. But often, scant regard has been
given to the ways in which the adoption of sound changes is mediated by the
This article enlarges on material presented at ICPhS ’99, San Francisco (Watt, 1999). The following
people deserve acknowledgment for their helpful advice and input: Gerry Docherty, Paul Foulkes,
Lesley Milroy, James Milroy, John Local, Joan Beal, Sandrine Dalban, Ann Williams, and Paul Ker-
swill. Several anonymous reviewers made valuable comments and suggestions. David Walshaw was
of great assistance with statistical matters, and I am grateful also to Penny Oxley, who carried out the
original fieldwork. The research was supported by the UK Economic & Social Research Council
(R00429524350).
Language Variation and Change, 12 (2000), 69–101. Printed in the U.S.A.
© 2000 Cambridge University Press 0954-3945000 $9.50
69

(unconscious) evaluative judgments of the speakers who are responsible for their
transmission.
The vowel system in such accounts has been conceptualized from the struc-
turalist notion of phonology as a set of oppositions, in which contrast is the key
factor (e.g., Trubetzkoy, 1969). The system is equipped with safeguards to its
integrity where contrast is threatened—say, where two contrastive categories be-
gin to approach each other, risking merger—and is configured to prevent the
development of too many contrasts (through split or borrowing). The vowel sys-
tem is thus formalized as a closed, self-referential, self-regulating submodule of
a language’s phonology. The principles by which vowel systems organize them-
selves are thought to be well understood and are uncontroversial in the sense that
they tend to be generally assumed in phonetics, phonology, and research into
language universals (Crothers, 1978; Disner, 1980; Liljencrants & Lindblom,
1972;Schwartz,Boë,Vallée, &Abry,1997; Vallée,Schwartz, &Escudier, 1999).
For instance, it is accepted that if a language’s vowel system features three con-
trastive categories (usually phonemes), these categories will be arranged in a
triangular configuration at the extremes of the possible vowel space; as more
contrastive categories are added, the overall triangular shape is preserved, with
vowels spacing themselves evenly along the periphery of the space. Throughout,
a pressure to maintain bilateral symmetry—by matching pairs of front and back
vowels at equivalent “heights”—is taken as a guiding principle. This last idea is
of particular relevance to the study presented here, and we return to it presently.
These principles continue to provide the basis for the conception of the vowel
system underpinning much of the recent work on linguistic variation (see, in par-
ticular, Labov, 1994), albeit in a somewhat diluted form. The fact that languages
havedialects andsociolects atallsuggeststhatthe independenceof thesystemmust
be moderated by the communicative and social needs of the speakers who use it,
sinceaspectsof thesystemcan bemodifiedfor thepurposes of stylisticand social
marking.Communities ofspeakers cantoasignificantdegreeacceleratethe spread
of an innovation or halt it in its tracks; according to Lennig (1978), the direction
ofsome vowelchanges apparentlymayevenbereversedwhilethey arein progress.
Alternatively, speakers may avoid the adoption of innovative forms altogether if
these are evaluated negatively.
2
The pressures on speakers and system should be
seen as complementary: any attempt to explain sound change must take both
speaker-centered(external) factorsand system-centered(internal)factorsintoac-
count. Asocially realistic model of vowel system dynamics must therefore allow
for a balance to be struck between the autonomy of speakers on the one hand and
that of the system on the other, since the relationship between speaker and system
is better viewed as symbiotic than as antagonistic.
The findings of the present study are interpreted as illustrative of this inter-
dependence. In what follows, the distribution of phonetic variants of the 0e:0
(face) and0o:0(goat) vowelsacross a sampleof 32 speakersof Tyneside(New-
castle) English (TE) is examined, with specific attention paid to the correspon-
dences of patterns within the distribution of their phonetic exponents with the
70 DOMINIC J.L. WATT

sex, age, and social class of these speakers and the phonological contexts in
which the phonetic forms occur. The vowel variables are referred to throughout
using Wells’s mnemonic keywords denoting lexical sets (Wells, 1982:xviii), thus
avoiding the customary, but sometimes confusing, use of phonetic symbols to
denote both speech sounds and abstract phonological categories. Modifications
to patterns in the data as a function of speaking style are also investigated.
Next, the data are appraised in terms of an internalist model. The question of
the relationship between face and goat is examined with reference to the struc-
tural concept of symmetry, as the face and goat vowels (or 0e0and 0o0) are
typically paired with one another in systemically oriented analyses of vowel vari-
ation and change. Since the present study is based upon a sample of speech drawn
fromalarge sampleofspeakers, however,the influenceofexternal, social factors
must also be given consideration. The patterns seen in the phonetic expression of
the face and goat vowels are amenable to analysis according to a dialect lev-
eling model, which would appear to capture the essence of these patterns rather
well. It is concluded that the data are better handled in terms of a dialect leveling
framework than by reference to the currently influential chain-shift model.
Before examining the data, some previous descriptions of the face and goat
vowels in TE are briefly discussed as a means of estimating the extent and direc-
tion of changes that have occurred over the last century or so.
FACE
AND
GOAT
IN TE
Newcastle upon Tyne, England’s northernmost city, is in economic and cultural
terms the hub of both the Tyneside conurbation and the far north of England (see
Figure 1). The dialect of English traditionally spoken on Tyneside retains a num-
ber of phonological features from Northumbrian Middle English (e.g., unshifted
[hus]forhouse, [nit]fornight, etc.)andin thisrespect is similarto Lowland Scots
and Scottish English (see Beal, 1993; Milroy, 1995). The persistence of these
features might be ascribed to Newcastle’s geographical proximity to the Scottish
urban centers and the continuous influx of migrants from Scotland over the last
few centuries (Beal 1993; Mess, 1928). On the other hand, the phonology of TE
conforms in other respects more closely to varieties of northern England (no
foot ;strut split, non-rhoticity, etc.; see Watt & Milroy, 1999). Given the
findings of recent studies of the consonantal aspects of Tyneside phonology (e.g.,
Docherty &Foulkes, 1999;Docherty,Foulkes, Milroy, Milroy,&Walshaw,1997),
it appears that the influence of southern English may be gaining ground, in that
there are signs of the adoption by Tyneside speakers of features such as labio-
dental variants of (r) and (th)-fronting,
3
which are thought to originate in south-
eastern England.
The face and goat vowels in TE are phonetically highly variable, as is clear
from previous literature on the subject (summarized in Table 1). Moreover, these
vowels appear to be undergoing change: comparison of speech samples for older
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 71

and younger TE speakers reveals a number of interesting differences between
these age groups. The hypothesized change could be characterized by an increas-
ing reduction in the use of forms specific to the Tyneside region accompanied by
the adoption of less regionally marked, supralocal forms. This process can be
seen as an aspect of the leveling of TE with respect to other forms of British
English. The results of the present study are fairly similar to those emerging from
research being carried out elsewhere in the United Kingdom, much of which
suggests that the sound changes underway in dialects of English around the coun-
figure 1. Location of Newcastle upon Tyne. Reproduced from Ordnance Survey maps
by permission of Ordnance Survey on behalf of The Controller of Her Majesty’s Statio-
nery Office, © Crown Copyright NC0000888.
TABLE 1. Variants of face and goat in Tyneside English,
as described in previous accounts
face goat
Jones (1911) [jE][e:i] [jE][o(:)u] [a:]
O’Connor (1947) [e:][e@] [o:][Õ@][Õ]
Viereck (1966) [e;][e:][e@][iE][jE][Õ:][Õ:][Õ@][iE][o:]
Hughes and Trudgill (1979) [e:][Ie] [Õ:] [uo] [a:] [e]
Wells (1982) [e(:)] [e@][Iá][o(:)] [e(:)] [U@][Õ(:)]
[Õ@]
72 DOMINIC J.L. WATT

try are part of a broad convergence of localized varieties on less localized ones
(for an overview, see Foulkes & Docherty, 1999).
Leveling of the vowel system of TE seems to have been underway for some
time. Viereck (1968), for instance, whose study of the TE of Gateshead focused
exclusively on vowel variation, asserted that
Some dialectal features seem to be more stable and less likely to succumb to Stan-
dard English in the near future. Others, however, will no doubt soon be completely
replaced, especially since the area under investigation is urban and consequently
the pressure of the standard language rather great, so that the traditional dialect is
bound to become increasingly mixed. Further, the fluctuation of the population
must be reckoned with as well as sociological factors, all of which contribute to a
dilution of traditional dialects....[T]he time will soon come when historically de-
veloped, genuine dialect phonemes are no longer heard.... All this, we feel, in-
creases the urgency of studying archaic, traditional dialect before these features
disappear completely. (Viereck, 1968:76)
While one might wish to take issue with Viereck’s claim that TE is converging
on (or “succumbing to”) Standard English, the general picture is very clear: the
high degree of dialect contact brought about by the “great number of people
[who] have poured into this area from other parts of the country” and the “various
influences[education,radio, television,film]which undermine[TE’s]traditional
character” (Viereck, 1968: 65) have in combination served to eradicate the use of
localized speech forms, replacing them with forms more typical of the English of
other parts of the country. Thus, we may expect that the face and goat vowels
under investigation here would not be exempt from this process of convergence,
although they could of course differ in the degree to which they are affected.
For either vowel, previous commentators (Hughes & Trudgill, 1979; Jones,
1911; O’Connor, 1947; Viereck, 1966; Wells, 1982) have identified a large range
of phonetic exponents, as specified in Table 1. It has been observed in the more
recent accounts (Hughes & Trudgill, 1979; Wells, 1982)—at least implicitly, in
that the vowels are described together in both works—that face and goat in TE
areina sense“mirrorimages” ofone another inaccents ofEnglish.Such apattern
among the range of variants listed in Table 1 is, however, perhaps less than im-
mediately obvious, given the range of attested qualities.
Since Jones’s time, the lexical membership of the face and goat sets for
manyTEspeakers hascome to approximatethat of southernBritish English more
closely; for instance, the neutralization of the face ;goat contrast at [jE]or[e]
in words like [tjEk] take and [bjET]both was—and still is, where it occurs—
restricted to a small number of words in the goat set. As shown by Jones and by
Hughes and Trudgill, [a:] can be found in TE goat words such as cold,snow, and
know, although this is an increasingly recessive feature.
On the basis of these five descriptions—weighted in favor of the more recent
studies of Hughes and Trudgill and of Wells—and from preliminary transcrip-
tions of contemporary TE made on the basis of the recordings used in this study,
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 73

three main types of variant are proposed for either vowel. These are listed in
Table 2.
The Type I monophthongs [e:] and [o:] are broadly typical of accents of north-
ern England and of Scotland (Wells, 1982) and thus are relatively unmarked in
TE with respect to neighboring varieties. Just as previous commentators like
Hughes and Trudgill or Wells used these symbols somewhat loosely—after all, it
would be unrealistic to expect every speaker of the variety always to produce [e:]
or[o:]with the cardinalvowel qualities thesesymbols represent inthe IPA—they
are used here to represent a restricted class of sounds: namely, clearly monoph-
thongal vowels produced close to the front and back peripheries of the vowel
space (thus including, e.g., [I:] and [E:] for face and [U:] and [O:] for goat).
Since the alternations of interest in the present study involve the alternation of
monophthongs with diphthongs, these qualitative variations are subsumed into
the broader categories indicated by [e:] and [o:]. The fact that earlier descriptions
ofTE generallyagreeon adistinction betweenmonophthongsand centeringdiph-
thongs provides additional support for the use of such a classification in this
study.
4
Type II diphthongs are to be found in other northern counties of England, such
as Yorkshire and Derbyshire (see Orton & Barry, 1969–71; Orton & Halliday,
1963), but it is probably true that they are now somewhat rarer in these regions
than on Tyneside, where they are by all accounts a long-established, traditional
feature of the accent. Holmes (2000) found that a panel of 40 listeners from four
regions of England (Tyneside0northeast; other northern; midland; southern) cat-
egorized [I@] and [U@], recorded in carrier words by a phonetician from outside
the Tyneside area, as characteristic of northeastern English with high levels of
consistency (on the order of 60% to 70%); the exception to this pattern of agree-
ment was among the other northern group, whose responses were split between
eastmidland,northern Scotland,and no response.
5
Thelisteners’responses to the
Type I variants [e:] and [o:] were much more mixed.While responses across all
four groups indicated that these forms were identified more often as northeastern
than as characteristic of any other region, there was a comparably high no re-
sponse rate (as high as 45% for [e:] among the Tynesiders, who clearly did not
perceive Type I monophthongs to be especially typical of the speech of their
TABLE 2. Variants of TE face and goat collapsed
into variant types
face goat
Type I: monophthongs [e:] [o:]
Type II: centering diphthongs [I@][U@]
Type III: closing diphthongs [eI] [oU]
(1[Õ:])
74 DOMINIC J.L. WATT

home area). The other northern group were again rather noncommittal here. For
[o:], the judgments were split equally between northeast and northwest (both
28%; 20% no response), whereas [e:] was identified as northeastern 10% of the
time, as northwestern 25% of the time, and as unlocalizable nearly half the time
(43%noresponse).Thus,non-Tynesidenorthernlistenersdid notcategorize Type
I monophthongs as being from anywhere in particular, but classified them as
northern when they classified them at all, and, as one might expect, they returned
vanishingly low responses (3% at most) for the southeastern or Standard English
categories. The consistent association of Type II variants with Tyneside among
listeners from various regions of England suggests that these forms have become
stereotyped, a factor which chimes with Wells’s (1982:375) remark that [I@] and
[U@] are “nowadays rather old-fashioned.” We return to this issue in subsequent
sections.
Again, the symbols used here should not be taken to indicate a unique, exclu-
sive phonetic quality.These variants incorporate any clearly diphthongal vowel
with an offglide relatively more central or open than the nucleus. The actual
phoneticqualitiesof thenucleusand theoffglidemay varysomewhat(cf. Hughes
& Trudgill’s and Wells’s variants in Table 1).
We turn next to the Type III diphthongs [eI] and [oU]. Note that, although
closing diphthongs of this type figure conspicuously in Jones’s account, they are
not cited by O’Connor, Viereck, Hughes and Trudgill, or Wells in their descrip-
tions of TE. Given the wide geographical dispersion of [eI] and [oU] (see Wells
1982:192–194, 210–211), this could be because these authors felt no need to
mention them in their discussions of the specifics of Tyneside phonology, and we
might conclude that [eI] and [oU] have been a feature of TE for at least a century.
But we should be careful here. Jones’s text was in fact a transcription of sixteen
lines of a song (Ah Wish Yor Muther Wad Cum, Joseph Wilson, c. 1860) with no
accompanyingcommentarybeyond footnoted glossesof some dialectwords, and
he used length-based transcription conventions (e.g., [hi:]he, [bit] bit; [hu:s]
house,[gud] good), which may have obscured qualitative distinctions. Further-
more, there is a scarcity of face and goat tokens in the text, with just three
examples each of face ([tjEk] take,[@gjEn] against, [we:iz] ways) and goat
([ko:ulz] coals, [sou] so,[Tou] though).
6
The closing diphthong in ways fits with
Viereck’s assertion (1968:70) that TE has tended to preserve a distinction be-
tween the reflexes of Middle English 0ai0(including ways) and those of Middle
English 0a:0(like take), but it is not obvious why, if this is so, against (ME0ai0)
patterns with take in Jones’s transcription rather than withways. The appearance
of [ou] for goat in the transcription is more difficult to explain. But since (i) we
are given no indication by Jones of the source of his sample (we are not told
whether this is a transcription of any one individual speaker’s productions, let
alone his or her sex, age, or background), (ii) we do not know whether this is a
transcription of speech or song, and (iii) there are only three tokens each of either
vowel in the text, we must rely more heavily on the evidence from other, later
sources. These sources do not indicate the presence of closing diphthongs of the
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 75

[eI];[oU] sort in TE. Indeed, Wells made specific mention of Tyneside as one of
a number of local accents that have implemented Long Mid Diphthonging—the
development in what Wells called “polite usage” of [e:] and [o:] into [eI] and
[oU], “only variably or not at all” (1982:211).
7
Therefore, we must assume that
they are not a traditional feature of TE, and that Jones’s text is unrepresentative.
In any case, there is little doubt that Type III diphthongs are very much a feature
of contemporary TE, although as we will see they are still somewhat sporadic and
confined on the whole to the speech of middle-class informants. Unfortunately,
Holmes did not include these variants in her perceptual study.
The fourth variant of goat, symbolized by [Õ(:)] in Table 1, completes the list
of variants for this vowel and throughout this article is symbolized by [Õ:] (as in
Table 2). It is fronted or centralized with respect to theType I monophthong [o:]
and has apparently been established in TE for some time, given that it is attested
by O’Connor as early as the 1940s. It is possible that this variant is a reflex of the
frontrounded[Õ:], whichis stillextantas agoat pronunciation inruralNorthum-
berland (Krause, 1989; Lass, 1989; Rydland, 1995). Alternatively, the absence of
[Õ:] from Jones’s description might indicate that goat fronting is an innovative
feature of TE, and that it is unrelated to the phonetic form of the vowel in Tyne-
side’s hinterland. The latter interpretation would certainly tally with the results
presented here and more generally with a pattern of back-vowel fronting that is
reported for English in other parts of the British Isles, such as Hull (Williams &
Kerswill, 1999), Bradford (Watt & Tillotson, 1999), Reading and Milton Keynes
(Cheshire, Gillett, Kerswill, & Williams, 1999), and around the world (e.g., La-
bov, 1994; Lass, 1989, 1990; Luthin, 1987;Watson, Harrington, & Evans, 1998;
Wells, 1982).
The [a:] variant, while extremely salient on those occasions when it is used, is
not included in the analysis due to its rarity in contemporary TE and to the fact
that its lexical distribution is in any case rather restricted (see Beal, 1985).
DATA COLLECTION
Recordings
A large corpus of recordings of conversational speech was used for the present
study. The corpus was collected for the project on Phonological Variation and
Change in Contemporary Spoken British English (ESRC R000234892; for de-
tails, see Docherty et al., 1997; Docherty & Foulkes, 1999; Milroy, Milroy, Do-
cherty, Foulkes, & Walshaw, 1999; Watt & Milroy, 1999) and is comprised of
approximately 26 hours of recordings of conversational English. Atotal of 32 TE
speakers were recorded talking in self-selected pairs (siblings, spouses, or close
friends of the same sex) in sessions lasting around 45 minutes. Towards the end
of the recording session, speakers read a 150-item word list (see Appendix). The
speakers were subdivided by three social variables—sex, age, and social class—
resulting in a total of eight speaker groups, each containing four speakers.
76 DOMINIC J.L. WATT

The younger age group contained speakers of between 16 and 25 years of age;
the older age group consisted of 45- to 65-year-old speakers. The social class
distinction—working class (WC) versus lower middle class (MC)—was based
onplaceof residence.Thetwo housingareasof Newcastlechosenfor thestudy—
Chapelhouse(MC) and NewbigginHall (WC)—were judgedby thefieldworker,
a local of Tyneside, to be a reliable guide as to the socioeconomic characteristics
of their residents. Her judgments were confirmed by 1991 UK National Census
information on a range of indicators (e.g., car ownership, proportion of adults in
employment, educational attainment).
The speakers were recorded in their own homes using high-fidelity digital
audio equipment and were encouraged to talk freely on topics of their choice with
minimal input from the fieldworker. On the rare occasions that the conversation
flagged, the fieldworker would address the speakers directly, asking them ques-
tions about their work or reminiscences of the past. The speakers were relaxed
and unself-conscious and talked readily to one another.
Transcription
Fromtheconversational material, aminimum of 30tokens per vowelofface and
goat was required for each speaker. Since both vowels are relatively frequent in
spoken English, the requisite number of tokens for each speaker was in all cases
easilyobtained. Onlyvowels inmonosyllablesorinsyllables bearingprimary word
stresswere transcribed.The (word-internal)postvocalicphonologicalcontextwas
noted, and tokens were grouped according to the following gross categories:
V#: open syllable; vowel final
Vn:V1nasal
Vp:V1voiceless plosive or affricate
Vb:V1voiced plosive or affricate
Vs:V1voiceless fricative
Vz:V1voiced fricative
Vl:V1lateral
A ceiling of 10 tokens of individual items was imposed to avoid lexically or
phonologically conditioned skewing of the sample. From the word list material,
all relevant items (i.e., those featuring face and goat vowels in stressed posi-
tion) were transcribed. The effects of the shift in speaking style brought about by
the switch from free conversation to the word elicitation task is discussed later.
By way of verification, comparison of the transcriptions was made with a set
of 1,112 face tokens and 1,130 goat tokens drawn from the same corpus and
transcribed independently by Lesley Milroy. For both vowels, the correlations
between the transcriptions were highly significant (at p,+002!,indicating a
very close match. We can therefore be confident that the transcriptions upon
which this study is based reflect the distribution of the principal variants of face
and goat in TE.
8
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 77

RESULTS
The overall percentage scores for each phonetic variant of face and goat in free
conversation (henceforth, FC) style are summarized in Tables 3 and 4, broken
down by speaker group. Effects for the speaker variables (sex, age, class) and
following phonological context in the distribution of the variants of either vowel
wereinvestigatedusing log-linearmodels (see,e.g.,Rietveld &van Hout,1993).
9
FACE data
Variables. A number of interesting effects on the distribution of phonetic
variantsofthis vowelcanbe observedin relation tothe sex,age,and classfactors.
As Table 3 suggests, the general pattern is a strong preference for the Type I
monophthong [e:] among all speaker groups except older WC men. Speakers in
the latter group appear to prefer the Type II centering diphthong [I@]. This diph-
thong is avoided almost completely by female speakers. The use of the Type III
closing diphthong [eI] among the speakers is overall rather rare, appearing most
TABLE 3. Variants of face, all speaker groups, FC style (%)
Group [e:][I@] [eI]N
Older MC men 78.3 21.7 — 143
Younger MC men 73.1 14.5 12.4 145
Older MC women 90.9 2.6 6.5 153
Younger MC women 79.5 2.4 18.1 166
Older WC men 36.2 63.2 0.6 174
Younger WC men 61.5 35.9 2.6 192
Older WC women 92.6 7.4 — 121
Younger WC women 97.4 2.6 — 151
Note: older 545 to 65 years; younger 516 to 25 years.
TABLE 4. Variants of goat, all speaker groups, FC style (%)
Group [o:][U@] [oU][Õ:]N
Older MC men 72.6 12.0 — 15.4 175
Younger MC men 44.7 2.9 17.6 34.8 170
Older MC women 89.8 — 9.2 1.0 196
Younger MC women 73.7 2.9 19.9 3.5 171
Older WC men 31.6 36.2 1.7 30.5 174
Younger WC men 59.2 12.0 1.0 27.7 191
Older WC women 98.9 0.5 — 0.5 190
Younger WC women 99.5 — — 0.5 197
78 DOMINIC J.L. WATT

frequently in the speech of MC women (and, to an unexpected extent, in that of
younger MC men; this group’s unanticipatedly high use of the [Õ:] variant of
goat is discussed later). For the face sample as a whole, then, we might expect
to see a strong sex-based effect, since the use of the Type I monophthong is heavy
among female speakers but is more variable among male speakers. This predic-
tion is confirmed by the very highly significant effect found for the sex variable
~p,+0001!+The variants’ distributions are also found to vary, although to a
lesserdegree,with class;sex andclassconsidered incombination yieldp5+0001,
and there is, as anticipated, a strong effect for sex, age, and class ~p5.0076).
Figure 2 gives an indication of the relative distributions of the three face
variants across the eight speaker groups. The sex-based distinction is readily
apparent here, as is the interaction of sex with the age and class variables.
Context. Postvocalic context also appears to have a strong influence ~p5
.0001) on the distribution of one of the three face variants (i.e., [I@]), but the
effect probably depends upon this variant’s distribution in the samples for male
speakers, since [I@] is used very little by female speakers: of a total face sample
of 591 tokens, it is recorded just 21 times (or under 4%). Table 5 therefore shows
just the figures for male speakers.
figure 2. Comparison of distribution of variants Type I, II, and III of face, all speaker
groups, FC style (%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 79

The environment in which [I@] is most strongly favored in male speech is the
Vp context for younger men and the Vn context for older men.
10
The proportion
of [I@] in open syllables ~V#!such as day,say, and gray is generally relatively
low. Despite the marked context effect, then, it is difficult to account in phonetic
or phonological terms for the variations in the appearance of [I@] from one con-
text to another, since the favored postvocalic segments do not appear to represent
any obvious natural class. No lexically related distribution was detected.
Style. A shift in attentional focus from the content to the form of speakers’
utterances might affect the distributions of the variants described earlier in var-
ious important ways. It is telling in this regard that, on being presented with the
word list, one older WC speaker (who had until that point been talking in a nor-
mal, unself-conscious manner) asked the fieldworker, “Do you want us to say it
the way it is on there or the way we would normally say it in a Geordie accent?”
(“Geordie” being a term Tynesiders apply to themselves and their dialect). His
conversational partner (his wife) responded, “I’m just speaking the way I’d nor-
mally say it.” Clearly, these speakers are fully aware of style shifting as a phe-
nomenon,ofthe expectationsmadeof themasreaders, andoftheir ownlinguistic
habits in such a situation; they are able verbally to specify the difference between
a mode of pronunciation customarily used for a formal task like this and normal,
TE-accented speech. It is interesting that the husband believes the written forms
to represent more closely a formal spoken register—“the way it is on there”—
than they do his everyday pronunciations. This is, of course, not evidence that the
variables discussed in the present study are necessarily subject to any such shift,
or that TE speakers are aware of or can talk about the alternations in the two
vowels. But recall the responses of the Tyneside listeners in Holmes’s identifi-
cation study, who classified the localized Type II [I@] and [U@] as specifically
northeastern forms much more consistently (77% and 76%, respectively) than
they did [e:] (33%) and [o:] (47%). If style shifting of the sort implied by our
speaker’s remarks involves a reduction in the use of Geordie features, then we
might anticipate a drop in the frequency of the Type II variants as the level of
formality increases.
The word list which informants were requested to read following the conver-
sational sessions contained 13 face items: gate (twice), paint,fatal,later,hate,
TABLE 5. Frequency of occurrence of [I@]variant of face in four contexts,
male speaker groups only (%)
Group V# Vn Vp Vs N
Older MC 21.2 26.2 22.4 10.5 310143
Younger MC 10.8 8.1 25.7 0 210145
Older WC 40.0 78.7 67.3 72.7 1100174
Younger WC 19.2 42.9 50.8 20.0 690192
80 DOMINIC J.L. WATT

eighty,eight,apron,matron,made,may, and tables. The sample sizes for each
speaker group were consequently a good deal smaller than was the case in the FC
style. The size of individual speakers’ samples was in some cases reduced by
misreadings or omissions of certain target words or was augmented by voluntary
re-readings, resulting in a small degree of variation in the total number of face
tokens in word list (WL) style from group to group. Table 6 shows a breakdown
of the face figures by variant, expressed as percentages; following context was
not taken into account in the analysis of the word list material.
From Table 6 it is apparent that, once more, the Type I monophthong is the
overallpreferenceacross thespeaker groupsas awhole.Thisisperhaps unremark-
able, since Type I variants are unmarked with respect to variants of Types II and
III. Note, however, that among those speaker groups who use the Type III closing
diphthong [eI] the proportions of the variant are much greater than is the case in
FC style. This suggests that these speakers may consider [eI] to be more appro-
priate for the reading task than the other available variants. Figure 3 compares the
distribution of [eI] between WL and FC styles, and it is apparent that this variant
is more typically a feature of careful style in TE speech.
11
As might be anticipated, [eI] is used a good deal more among MC speakers
than among WC speakers. There is a suggestion of a sex bias here—specifically,
of an association of [eI] with female speech. Use among the WC groups is overall
rather low, with only older WC women using [eI] to any extent. Within the MC
groups, on the other hand, the distribution is complex: all four groups use [eI]at
least some of the time in WL style, and in all four cases its use in WL style is
greater—for older MC women much greater—than in FC style. The weak posi-
tive correlation between the data sets for WL and FC styles ~r50+428,p.+5!
confirms the lack of close identification between the distributions of [eI] in either
style, thus supporting the claim that this variant is heavily marked stylistically
and is perhaps considered more “correct” than the alternatives. The loss of defi-
nition in the sex distinction between the two MC age groups is once more indi-
cation of leveling.
The relative increase in the use of the Type II variant [I@] by WC men in WL
style (compare Tables 3 and 6) is not entirely expected, however. It may be that
TABLE 6. Variants of face, by speaker group, WL style (%)
Group [e:][I@][eI]N
Older MC men 84.6 — 15.4 52
Younger MC men 51.9 9.6 38.5 52
Older MC women 26.9 — 73.1 52
Younger MC women 73.6 — 26.4 53
Older WC men 26.4 73.6 — 53
Younger WC men 50.0 50.0 — 52
Older WC women 72.5 — 27.5 51
Younger WC women 98.1 — 1.9 54
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 81

these speakers are simply less sensitive to the pressures that cause female and0or
MC speakers to adjust their pronunciations in a direction away from the localized
forms. Or perhaps, following Veatch (1991), we might speculate that the center-
ingdiphthongrepresents amore emphatic,hyperarticulated pronunciationof what
is nominally a peripheral monophthong. In any event, it seems that style shifting
of the sort apparent in the speech of other groups is not in evidence in the use of
face variants among male WC speakers.
GOAT data
Variables. Recalling the figures shown in Table 4, we can observe a series of
parallels with face with respect to the socially conditioned patterning of the four
posited goat variants. The distribution is more easily visualized if displayed
graphically, as in Figure 4.
The general preference among the speakers in the present sample is once again
for the generic northern Type I variant [o:]. However, two groups diverge con-
spicuously from this pattern: older WC and younger MC men use it in just 31.6%
and 44.7% of their respective samples, preferring instead to use above-average
levels of the Type II diphthong [U@] and the fronted variant [Õ:]. To see such
indications of dialect loyalty among older WC men seems reasonable, but what is
less obvious is why younger MC men would use the [Õ:] variant more than any
other speaker group. Some possible interpretations of this behavior are offered
later on.
The second trend concerns the distributions of [U@] and [Õ:] versus the distri-
bution of [oU]. The first two forms appear to be in roughly complementary dis-
figure 3. Type III face variant [eI] inWL and FC styles, all speaker groups (%).
82 DOMINIC J.L. WATT

tributionwiththe third,inasmuch asspeakers whouse [U@] and0or [Õ:]are unlikely
to use [oU] to any great extent, and vice versa. [U@] and [Õ:] can thus be thought
of as male forms, whereas [oU] is a predominantly female form that is also pop-
ular to some degree among young MC men. Perhaps [oU] might be better char-
acterized as a form that is most frequent among MC speakers, particularly young
ones.
Statistical analysis of the data reveals that the effect of speaker sex upon the
distribution of the four goat variants is once again strongest ~p,.0001) where
the data are grouped to compare the pooled figures for the localized forms [U@]
and [Õ:] against the (distinct) sets of scores for [o:] and [oU].
12
This is also true of
a combined effect of sex and class ~p,+0001!,reflecting a tendency among both
MC men and women (WC speakers avoid the [oU] variant almost completely) to
prefer [oU] as a function of age.
Context. As was the case for face, a context-related effect seems tied most
strongly to the distribution of the Type II diphthong [U@]. Context in combination
with the sex–class effect has a very highly significant influence ~p5+0002!,
which is presumably—at least in part—a consequence of the virtual absence of
[U@] in female speech. Four of the six [U@] tokens recorded for female speakers
(of a goat total of 754 for the women) are found in the Vl context, which may be
related to the propensity of TE speakers to append a centering offglide to canon-
figure 4. goat variants, by speaker group, FC style (%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 83

ical monophthongs before [l], which in TE is typically clear in all positions (see
Viereck, 1966:69, for examples). Since [U@] is virtually unused by female speak-
ers, and since class is strongly implicated in the distributional pattern where [U@]
is used by male speakers, the figures for the older and younger male groups are
collapsed by class, as shown in Table 7.
Roughly speaking, there is a tendency for [U@] to be favored in the Vs,Vl,and
(for WC men) Vn positions; the variant is not recorded at all for the MC group in
the Vp and Vz contexts. This is hardly surprising, given the overall infrequency of
the variant among MC speakers. There is nowhere a consistent alternation that
suggests allophony as such. Nor could any influence of lexical identity on vowel
quality be detected. Once again, it appears that the distribution of the four vari-
ants of goat in an individual TE speaker’s sample depends more on the social
attributes of the speaker than on phonological or lexical conditioning.
Style. goat is represented in the word list by 8 items: boat (twice), total,
motor,wrote,load,go, and won’t. The figures recorded for goat in these items,
as shown in Table 8, show that a similar pattern obtains for this vowel as is the
case for face. [o:] is overall the most popular form, and the local variants [U@]
and [Õ:] are less widely used than in FC style. Once more, older WC men exhibit
no style shifting with respect to the latter forms. Type III diphthong [oU], like its
face counterpart [eI], is proportionately more common in WL style for the sam-
ple as a whole than is the case for FC style. The style-shifted distribution of [oU]
matches very well with that for [eI], as Figure 5 suggests; the lack of identity of
the FC and WL figures for goat is reflected by the correlation coefficient of
0.638 (Pearson’s r;p.+2!for these figures.
The tendency to substitute [oU] for other variants of goat is again greatest
among older MC women, even though this group uses [oU] less in FC style than
do younger MC men and women. We might conclude from this patterning that, of
all the speaker groups investigated here, older MC females make the strongest
equation between Type III diphthongs and “carefulness” of pronunciation, their
preference in unmonitored speech for the monophthongs [e:] and [o:] notwith-
standing. This pattern is reiterated among older WC women, but to a much less
dramatic degree. Atendency of this sort is not entirely surprising, given that the
[oU]variantmay wellbe considered aprestige variantbyolder speakers.AsWells
commented, “[ÆU] has only quite recently (since the Second World War?) ousted
TABLE 7. Frequency of occurrence of [U@]variant of goat in seven contexts,
male speaker groups only, by social class (%)
Group V# Vn Vp Vs Vb Vz Vl N
MC 4.5 5.8 0 14.3 8.7 0 17.5 260345
WC 13.6 30.8 4.8 21.6 36.4 35.1 29.4 860365
84 DOMINIC J.L. WATT

[oU], or perhaps rather [öU], as the ideal image of a ‘correct’ or ‘beautiful’ RP
goat diphthong” (1982:237). Such an evaluation may be an influence here.
Summary
The figures for the phonetic variants of face and goat in TE are fairly typical of
patterns reported in other studies of dialect leveling in British English, inasmuch
asthedecline of traditional,localized speech formsis balanced (orcaused) bythe
substitution of less marked forms typical of a broader area. The main points that
emerge are as follows.
TABLE 8. Variants of goat, by speaker group, WL style (%)
Group [o:][U@][oU][Õ:]N
Older MC men 90.6 — 9.4 — 32
Younger MC men 36.4 — 33.3 30.3 33
Older MC women 43.8 — 56.2 — 32
Younger MC women 71.9 — 28.1 — 32
Older WC men 3.1 65.6 — 31.3 32
Younger WC men 51.4 5.7 — 42.9 35
Older WC women 80.6 — 19.4 — 31
Younger WC women 96.9 — 3.1 — 32
figure 5. Type III goat variant [oU] inWL and FC styles, all speaker groups (%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 85

1. Type I monophthongs [e:] and [o:] are most frequent across the speaker group as
a whole.
2. Type II centering diphthongs [I@] and [U@] are common among maleWC speakers,
but appear to be becoming recessive.
3. Type III closing diphthongs [eI] and [oU] are on the increase, particularly among
younger MC speakers of both sexes.
4. Women appear to be leading the changes, and they are most advanced among MC
speakers.
5. The centralized variant [Õ:] is apparently in decline, but is used at high levels by
younger MC men.
6. The identity of the phonological context following the vowel is found to have a
strong effect on the distribution of the variants of either vowel, although there are
no patterns consistent enough from group to group to suggest an allophonic alter-
nation. Afollowing nasal appears weakly to favor the use of Type II diphthongs
among male speakers.
7. There is a marked difference between the distributions of face and goat variants
as a function of speaking style: specifically, the use of localized Type II forms
becomes almost negligible except among older WC men in WL style, whileType
IIIvariantsare favoredmore stronglyin WLstyle thanin FCstyle byother speaker
groups.
Given that face and goat are overall rather similar to one another in the
typological characteristics of their chief variants and in the distribution of these
variants among the TE-speaking population sampled here, it seems reasonable
now to compare them against one another directly. In the following section, the
similarity between face and goat in terms of the proportions of their phonetic
exponents across the eight speaker groups is considered, and the proposal that
face andgoat havesimilar surface formsbecause theyare phonologicallypaired
(or vice versa) is evaluated.
FACE
AND
GOAT
AS PARTNER VOWELS
Ithasbeen suggestedthatthe localizedType II formsare being lostfrom TE.That
is, we have evidence of a sound change in progress characterized by the rejection
of a traditional diphthongal form in favor of a supralocal, monophthongal form
and the adoption among certain speakers—young and0or female MC ones,
predominantly—of a supralocal diphthongal form more typical of varieties spo-
ken to the south of Tyneside. The reduction in the use of the traditional forms is
thereby balanced by the simultaneous adoption of supralocal forms, which are
less geographically (and perhaps socially) marked. In other words, the patterns in
the face and goat data are part of a process of dialect leveling, the hallmark of
this process being a situation whereby heterogeneous speech varieties over time
become more homogeneous, either by converging upon a pre-existing variety or
by coalescing into an entirely new one (see, e.g., Auer & Hinskens, 1996; Hin-
skens, 1998; Williams & Kerswill, 1999).
86 DOMINIC J.L. WATT

Whatissurprising inthepresent caseis the similarityof thepatternsin thedata
sets for face and goat when these are compared with one another and the ap-
parentunderlyingorderliness thissimilaritysuggests. Whena dialect issubject to
leveling, we might expect the system shared by the dialect’s speakers to be tem-
porarily thrown out of any equilibrium it might earlier have attained. One might
imagine periods of linguistic change resulting from dialect contact (rather than
the slow, gradual, internally choreographed reorganizations postulated by histor-
ical phonologists) to be characterized by some disorder and confusion at the
phonetic level and quite possibly at the phonological level as well. Yet, in the
following comparison of the face and goat figures, the pattern we see suggests
that these vowels are leveling, as it were, in lockstep with one another.
Type I monophthongs
The percentage scores for each speaker group’s use of theType I monophthongs
[e:] and [o:] are shown graphically in Figure 6. The relative proportions of Type
I variants in the samples for each speaker group can be seen to parallel one an-
other rather closely.The correlation (Pearson’s r!between the data sets for face
and goat was again assessed on the basis of the scores represented in Figures 2
and 4. The visual match in Figure 6 is borne out by the strong positive correlation
between face and goat for this variant type ~r50+917,p,+002!+Only among
figure 6. Comparison of Type I variants of face and goat, all speaker groups (%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 87

younger MC male speakers does this trend seem to diverge to any extent. For this
group, a higher than average proportion of the sample is accounted for by the [Õ:]
variant of face; indeed, as we have seen, this group uses [Õ:] more than any other
speaker group. Overall, though, there is a remarkably close fit between the Type I
figures for both vowels.
Type II diphthongs
The agreement between the percentages for Type II variants is also strong ~r5
0+964,p,+002!+As can be seen from Figure 7, there is some divergence in the
levels of usage of Type II face and goat variants among male WC speakers;
again, this is probably the consequence of the presence of a high proportion of
[Õ:] in the samples for these speaker groups. The overall proportions of Type II
diphthongal variants in the samples for the two vowels nonetheless approximate
each other.
Type III diphthongs
The correlation ~r50+981,p,+002!between the percentage scores for Type III
closing diphthongal variants of face and goat is evident in Figure 8. MC
speakers—most particularly, the younger ones—are seen to use [eI] and [oU]
relatively frequently by comparison with other speaker groups. Among older MC
men and the entire WC cohort, the use of Type III variants is negligible.
figure 7. Comparison of Type II variants of face and goat, all speaker groups, FC
style (%).
88 DOMINIC J.L. WATT

The strength of the correlation in the case of Type III variants and the fact that
these forms are avoided altogether by certain speaker groups indicate that the
adoption of [eI] and [oU] among MC speakers is highly socially marked. These
forms are more typical of varieties of English spoken to the south of Tyneside,
including status forms such as Received Pronunciation, and have been associated
with “polite usage” by Wells (1982:211). The overall proportions of these vari-
ants are, of course, still fairly small, but this is to be expected if, as was argued
earlier, they are indeed a relatively new feature of TE.
The parallelism between face and goat in FC style may be just as apparent in
WL style. That is, the effect of a change in style in one vowel may be very similar
in the other. Such style shifting is examined next.
Style shifting
As was suggested earlier, the direction of the style shifting evident for the face
and goat vowels appears to be rather similar: Type I monophthongs are (as in FC
style) the most popular variant overall in WL style, whereas the Type II variants
are used hardly at all except by older WC men. Type III variants of both vowels
are used significantly more often in WL style than in FC style.
Theclosecorrespondences betweenthe pairedpercentagescores foreach vari-
ant in the WL style for face and goat are evident in Table 9 (this excludes [Õ:],
figure 8. Comparison of Type III variants of face and goat, all speaker groups, FC
style (%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 89

which has no parallel form in face). Indeed, the rvalues for each pair of variants
suggest that variants of Types I, II, and III of face and goat are almost as closely
matched from speaker group to speaker group in WL style, as is the case in FC
style.
Again, the presence of [Õ:] in the samples for the three groups of the male
speakers (younger MC, older WC, younger WC) skews the goat sample some-
what, and there are disparities in the Type III samples for several groups. But
thesearenot sufficientto affecttheoverall patternof similarity betweenface and
goat. Hence, the parallelism of these two vowels, as expressed by their various
phonetic exponents, is manifest also when the speakers’ productions are subject
to the higher level of self-monitoring typical of word list reading tasks.
In the following section the nature of the parallelism is explored in more de-
tail. Specifically, the question is addressed as to whether we can attribute the
similarities between face and goat at the surface level to an underlying sym-
metrical relationship between the vowels—a relationship which would almost
certainly be assumed in an account focusing on the contribution of system-
internal forces in the course of a sound change.
INTERNAL FACTORS
It has been argued that the conception of the vowel system that is current in
sociophonology is essentially the same as that inherited from the structuralist
tradition. The vowel system is, in such a model, a set of categories that are kept in
opposition with one another through the operation of pressures which (ideally)
serve to maximize the acoustic (and0or perceptual) distance between the catego-
TABLE 9. Comparison of variants of face and goat, excluding [Õ:],
by speaker group, WL style (%)
Group [e:] [o:][I@][U@] [eI] [oU]
Older MC men 84.6 90.6 — — 15.4 9.4
Younger MC men 51.9 36.4 9.6 — 38.5 33.3
Older MC women 26.9 43.8 — — 73.1 56.2
Younger MC women 73.6 71.9 — — 26.4 28.1
Older WC men 26.4 3.1 73.6 65.6 — —
Younger WC men 50.0 51.4 50.0 5.7 — —
Older WC women 72.5 80.6 — — 27.5 19.4
Younger WC women 98.1 96.9 — — 1.9 3.1
N254 153 70 23 95 48
r0.917 0.848 0.986
p,.002 ,.01 ,.002
Overall: r50+933,p,+002
90 DOMINIC J.L. WATT

ries. Vowels thus locate themselves in vowel space in an orderly and quasi-
predictableway,in thesensethat, fora givennumber oflexical contrastsexpressed
by the vowel system (i.e., vowel phonemes), there appears to be only a fairly
small number of configurations attested cross-linguistically. One of the guiding
principles at work here is that languages prefer symmetrical vowel systems; this
being the case, the range of possible configurations from which a system of con-
trasts of a given size can choose is rather constrained.
The idea that 0e0and0o0(TEface and goat) are symmetrical partner vowels
is implicit in such a model; typically they are described as being of equivalent
height and tenseness. This is assumed at an abstract phonological level, but is
claimed (against a rather heavy body of articulatory, acoustic, and perceptual
evidence, it has to be said) to be true also at the phonetic level, even when change
is underway. Pfalz (1918), for instance, described vowels as marching in rows
(Reihenschritte) and presented a table of correspondences between the phonetic
forms of 0e0and 0o0in Germanic languages; he asserted that the phonetic form
of 0e0determines that of 0o0in a given language or dialect (and presumably vice
versa).Ofsignificancetothe presentstudy ishis statementthat “Wherethee-sound
is )e@, the o-sound will be )o@” (1918:28, cited in Stewart, 1976:87; my transla-
tion). The Type II diphthongs in TE would appear to conform to this axiom, as
would the other face and goat variants, with the exception of [Õ:].
The actual origin of Type II diphthongs in other varieties of English for which
they are reported (Northern Ireland; Jamaican; Fond du Lac, Wisconsin) is ac-
counted for by Veatch (1991:186) as the consequence of face and goat being
strongly stressed: that is, the raising and breaking of [e:]to[I@] and [o:]to[U@]
result from hyperarticulation (cf. Kerswill, 1984:25–26, who found a strong cor-
relationbetweenprimary stressandthe appearanceofa centeringdiphthongvari-
ant of face in Durham English; see also Samuels, 1972:21–27). Labov,Yaeger,
andSteiner(1972:97)brieflydiscussed the socialdistribution of [i;@]and [u;@]in
the English of Gateshead, citing the appearance of these forms as an example of
the historically common process of raising tense ingliding vowels (see especially
Chapter 3), but they made no attempt to provide an explanation of this develop-
ment in articulatory phonetic terms. The diphthongal forms are instead the prod-
uct of gradual drift. Labov et al. presented evidence to show that the raising and
breaking of these vowels are historically common among Germanic, Romance,
Balto-Slavic, and Semitic languages: in fact they used the term “symmetrical” to
describe the shift, which they schematized as follows.
In Veatch’s account and in that of Labov et al., the coexistence of Type I
monophthongs and Type II diphthongs is the result of the development of Type II
diphthongs from [e:] and [o:](Se and So in Labov et al.). Furthermore, the appear-
ance of Type III diphthongs [eI] and [oU] is accounted for in Labov et al. as the
final stage of the same shift.
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 91

When the [raising and breaking of Se and So] is completed, and the nucleus reaches
high position, the next step is usually either monophthongization or a shift to a
rising diphthong. (Labov et al., 1972:104)
As they stand, the patterns evident in the TE data might exemplify the three
stages of this shift quite well. [e:] and [o:], it could be argued, raised in tandem
and both developed centering offglides ([I@] and [U@], respectively); later these
centering diphthongs somehow shifted to [eI] and [oU]. Presumably, however,
these shifts did not affect both vowels across the board, since all three types of
variant are still extant for face and goat: that is, the shift from Type I monoph-
thongs to Type II diphthongs was not completed before Type III diphthongs ap-
peared on the scene. The system of alternants described earlier could then be seen
as a “fossilization” of a change that had been arrested by some means. Alterna-
tively, the patterns in the contemporary TE data might indicate that the sound
change is in progress and is as yet incomplete. Throughout, however, the theme
of symmetry is implicit: whatever is true of one vowel will be true of the other.
The provenance of [Õ:] is also unproblematic if several forms of a vowel can
coexist in this way. Fronting of back vowels is a commonly reported type of
sound change, not least in English, and indeed Labov (1991, 1994) provided a
principle to account for shifts of this type: “In chain shifts, back vowels move to
the front.” So although the appearance of [Õ:] disrupts the neat symmetry of the
paired variants of Types I, II, and III, it can still be accounted for under the
chain-shift model, and in spite of its presence the symmetrical pattern obtaining
between face and goat is still readily apparent.
The results of this study might therefore be taken as good evidence of inter-
nallymotivatedshifting intheTE vowelsystem.The principleof symmetry isthe
crucial factor conditioning the path of the change: it serves to keep changes in
both vowels in line with one another. There are, however, a number of problems
with this explanation. If the shift is unidirectional ([Type I] ][Type II] ][Type
III]) (i.e., if the appearance of Type III diphthongs depends upon the adoption of
Type II diphthongs), it is difficult to account for the relatively high frequency of
Type I monophthongs over the other two types when we bring their distributions
across the speaker sample into the picture. If variants of Type II develop from
Type I variants, why are the former apparently recessive, being favored by older
WC men but virtually absent among female speakers? Given what we know
about the adoption of innovatory forms in English-speaking communities, we
would expect the opposite to be true. Type III variants [eI] and [oU] are, of course,
most widely used by women, but in order for them to arrive at that stage of the
shift they would have had to pass through the intervening Type II variant stage,
according to Labov et al. Apparently, this is not the case, since where female
speakers do not use the Type III closing diphthongs they stick resolutely to the
TypeImonophthongs.One wouldanticipate Type IIIdiphthong usageto beheavi-
est among WC men because they use Type II diphthongs [I@] and [U@] more
frequently than do the other groups and are thus, by Labov et al.’s criteria, fur-
92 DOMINIC J.L. WATT

thest along the track of the change. But, as the results presented earlier show, this
is clearly not the case.
What is missing here is some account of the social marking that is attached to
each of the variants. In other words, the patterns seen in the data are not inter-
pretable entirely in terms of alterations the system makes to itself. Rather, we
need to take into consideration the probability that the source of the Type I and III
variants is not internal but external, and that these and the other variants of face
and goat are ascribed values on the basis of (and that determine) their distribu-
tion among the TE speakers’sample. In the following section, the effects of some
external factors are discussed, and the question of dialect leveling is raised, since
in combination these account more plausibly for the variation in TE face and
goat than does the internalist model on its own.
EXTERNAL ACCOUNTS
Consider the pattern in Figure 9. What is shown here is a comparison of the male
speaker groups with respect to the proportions of Type II variants [I@] and [U@]
and the fronted goat variant [Õ:] in their samples. The samples for the female
speaker groups, being negligibly small, are omitted from the figure.
figure 9. Proportions ofTypeII and[Õ:]variants, malespeakergroupsonly,FC style(%).
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 93

Except for the case of [Õ:] among younger MC men, the trend suggested by
Figure 9 is one of a general decline in the use of all three variants as a function of
age,thisdecline beingmore advancedamongMC speakers.These variantsare,as
has already been pointed out, more localized to Tyneside than are Type I and III
variants, and so the use of [I@], [U@], and [Õ:] would be more strongly indicative
of local identity than would the use of other forms. As with nonstandard phonol-
ogy elsewhere in the United Kingdom, a certain degree of stigma may well be
attached to their use; the direction of the style shifts described earlier would
imply this, at any rate. What Figure 9 suggests, then, is that use of Type II vari-
ants is strongest among older WC men because it symbolizes local affiliation (as
Holmes’s results suggest) and also perhaps loyalty to traditional values and is
lowest among younger MC men because of their relatively weak identification
with these values. The sensitivity of younger MC men to the markedness of Type
II diphthongs appears to be sufficient to suppress the use of these forms almost
completely in the careful WL style, as indicated in Tables 6 and 8; older WC men,
on the other hand, actually increase their use of the centering diphthongs in WL
style,indicatingthat theyevaluate theseforms quitedifferentlyfromothergroups.
The path of the change is thus arguably determined by the acceptance or rejection
of Type II forms on the basis of how speakers evaluate them.
The adoption of Type III diphthongs into the TE face and goat repertoire
among women and younger MC speakers would suggest the same conclusion:
[eI] and [oU] are more characteristic of accents used to the south of Tyneside than
they are of TE itself and therefore may be evaluated as more attractive than the
local options by these speakers, although it is probably true to say that such
perceptions have little to do with the geographical origin of these variants; as we
saw earlier, Wells located their origin in “polite English usage,” an association
which may persist.
Across the sample as a whole, however, the preference is overwhelmingly for
Type I monophthongs for both vowels and in both FC and WL styles. Whether
these were the original input to the shift that produced the raised and broken Type
II variants is in a sense immaterial, since the Type II variants [I@] and [U@] appear
to be declining at the expense of [e:] and [o:], regardless of which preceded the
other. Rather than postulate a reversal of the shift, as a purely internal account
might demand, it seems more plausible to explain this increase in terms of the
relationship between TE and neighboring varieties of northern British English
with respect to these monophthongs. [e:] and [o:] are, as Wells (1982:364–365)
and others pointed out, generally typical of English in northern England and, as
Holmes (2000) indicated, are thus marked for northernness but not for locality
more narrowly than this. It will also be recalled that Wells judged the Type II
diphthongs to be rather old-fashioned. Presumably, then, the increase in Type I
monophthongsasa generalfeatureofTEcan beseenas convergenceon abroader
regional pattern, as one might expect where leveling is taking place. Type I mon-
ophthongs are under this interpretation acceptable to most TE speakers because
they are less likely to provoke in listeners any negative stereotyping associated
with the more specifically northeastern [I@] and [U@].
94 DOMINIC J.L. WATT

Considering next the distribution of [Õ:], however, we can see from Figure 9
that the reduction of the proportions of this variant among older WC, younger
WC, and older MC men is countered by younger MC men, who use [Õ:] more
than any other group. This finding tallies with Wells’s observation that “the cen-
tral rounded monophthong [Õ(:)] remains a very characteristic goat quality both
for Tyneside itself and for all Northumberland” (1982:375). This should be qual-
ified, of course, with the remark that [Õ:] is not at all characteristic of the speech
of the Tyneside women in the present sample, and, apart from the resurgence of
this form among younger MC men, it could be said to be in decline overall in TE.
Here, we might consider the revitalization of [Õ:] among younger MC men to
parallel the unexpectedly high frequency of local fronted forms of 0ay0and 0aw0
in the speech of young, educated men on Martha’s Vineyard, a pattern Labov
(1963) interpreted as a reaction among these speakers to the encroachment of
variants from the U.S. mainland. In much the same way,TE [Õ:] could be thought
of as a form which younger MC men consider attractive in the sense that it is
recognizably northeastern and a variant that does not suffer to a comparable ex-
tent from the stigma of old-fashionedness that may be attached to [U@]. As we
haveseen,[U@] isaform youngerMCmen avoidalmostcategorically inFCstyle,
and it is altogether absent in WL style for this group.
If the use of [Õ:] counts as goat fronting, it could be maintained that La-
bov’s Principle III (“In a chain shift, back vowels move to the front”) applies
here. However, the operation of this principle in Labov’s model is confined to
vowel nuclei that are fronting along peripheral tracks (i.e., along the upper
0i0;0u0continuum or along the 0a0;0A0continuum). Fronting of goat, La-
bov contended, entails a coordinated shift with 0u0(i.e., the goose vowel),
whereby goat fronting “represents a generalization of the fronting of the high
back vowel ....When 0ow0is fronted, it is always in parallel with 0uw0and
considerably behind it” (1994:208).
13
Fronting of goose, however, while a fea-
ture reported for many varieties of British English, is not described in any
previous accounts of TE and is not found with any regularity in the TE record-
ings used for this study. If anything, the quality of TE goose is generally as
close to the close back rounded vowel represented by cardinal vowel 8 as can
be found in any variety of modern spoken English, at least in checked syllables
like boot,book,food,lose, and so on (see Watt & Milroy, 1999). Bearing in
mind the fact that [Õ:] is reported as a goat variant as early as O’Connor
(1947), goose fronting ought to be an established feature of the dialect if La-
bov’s stipulation is to hold for TE. Since apparently it is not, we must question
the relevance of Principle III here or, at any rate, the goose fronting precondi-
tion; without the goat fronting ]goose fronting implication, though, Princi-
ple III seems to lose force as an explanatory principle. The influence of external
factors—namely, that younger MC men are expressing a preference for a pre-
existing localized variant they adopt as an assertion of local identity—provides
us with a more coherent explanation for the increase in the use of [Õ:]. (Com-
pare this with the high levels of use of [Õ:]asagoat variant among middle-
class Hull girls described in Cheshire et al., 1999.)
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 95

CONCLUSIONS
Overall, the patterns of usage of the phonetic variants of these two vowels are
much more complex than would be implied by an analysis focusing on the rela-
tionship between face and goat at an abstract phonological level. In order to
arrive at an understanding of the forces at work behind the distributions of the
described variants, we must consider not only the possible origins of each form,
but also the significance these forms hold for the speakers who use them and the
degree to which they are used in speech as a reflection of the social structure of
the community in question.
While the correspondences between the individual variants of face and goat
are indeed close and thus strongly suggestive of underlying structural symmetry,
we might also take the view that, for example, [I@] and [U@] pattern alike because
they are equivalently socially marked. If both are considered old-fashioned, in-
dicative of an inward-looking, strongly locally oriented attitude, or incorrect,
they are on balance more likely to be avoided by women and middle-class speak-
ers, who have been shown in numerous studies to disfavor variants associated
with these traits. The same may be said of [eI] and [oU]: if TE speakers associate
these with correctness, as the style shifts described earlier suggest, it is probable
thattheywillfindgreater favoramong younger,middle-classspeakersand women.
The unmarked variants [e:] and [o:] would then fill in the gaps, acting as default
variants lacking strong regional or social marking in either direction. This being
the case, an apparently symmetrical pattern may start to emerge.
Also, we should be careful about generalizing any such symmetrical pattern to
other pairs of vowel categories (e.g., fleece and goose) without examining their
surface formsata levelof detailsuchas thatused hereforface and goat;fleece
andgoose inTEare infactfound tosharecertainsimilarities withoneanother with
respect to an allophonic alternation between the vowels in open and checked syl-
lables(Watt&Milroy,1999),butthisis difficultto integratewiththeobserved pat-
ternsinface and goat inany waycommensurate withthechain-shift model.And
whyis symmetryconsidered astrongguidingprinciplein vowelsystemsbutnotin
consonantsystems? Pairingssuchas0p0and0b0or 0s0and0z0areoftendescribed
asbeingofa similartypeto, say,0i0and 0u0or0E0and0O0,butwhere soundchange
affects one category it is not immediately assumed (as per Pfalz’s Reihenschritte)
that the change will equally and simultaneously affect its partner.
14
In structural
analysesofconsonant systems(Hockett, 1958;Samuels,1972), symmetryis pos-
tulated as a design feature, but implicit in these is the suggestion that our reasons
forconceivingof consonant systemsas symmetrical isas much motivatedby aes-
thetic considerations as by linguistic ones (cf. Hockett’s principle of neatness of
pattern,exemplifiedby hisinstruction that“if weareconfronted withtwo ormore
waysofidentifying allophones asphonemes, both orall of whichequally meet all
othercriteria,weshouldchoosethatalternativewhichyieldsthemostsymmetrical
portrayal of the system”; 1958:109, quoted in Stewart, 1976:85).
15
We should note that studies of socially conditioned variation in consonantal
variables sometimes report patterns that make the variables appear to be linked to
96 DOMINIC J.L. WATT

one another in much the same way as was found here for face and goat. James
Milroy (1996), for instance, compared the distributions of labiodental variants of
(th) ([fIÎ] for thing, [i:v@] for either, etc.) with those of pre-vocalic and pre-
pausal glottal stop variants of (t) in Derby English. Ranking his eight informants
according to their respective use of labiodentals and glottal stops revealed a cor-
respondence between proportions of labiodental (th) and glottaled pre-vocalic (t)
thatwasstatistically significant.Butno plausiblephonological connection canbe
drawn between these two variants. Rather, it might be speculated that their dis-
tributionsaresimilar foreach oftheDerby speakerssampled becausethese speak-
ers perceive labiodental (th) and pre-vocalic [?] to have approximately similar
attributes with respect to social marking, and they are thus adopted to a greater
and lesser degree on the basis of their appropriateness to the self-identity of each
speaker (see also Thelander, 1982, for relevant discussion).
In conclusion, the changes described in this article may be understood as
speaker-motivated adoptions and rejections of sociolinguistically marked sur-
face forms brought about by differentials in social psychological attitudes, per-
haps reflecting factors such as local loyalty, a shift in the balance of identity with
respecttobroader regionratherthan immediatelocation,or somedesire to appear
modern, educated, or well-spoken. These speech forms are either pre-existing in
the variety or are available by borrowing from other neighboring varieties. No
appeal is necessary to spontaneous creation by the internal workings of the vowel
system, even if these were responsible for the origin of each form in the first
place. In this case, an approach which emphasizes the social motivations for
modifications to the phonetic expression of phonological categories is preferred
to an analysis which attempts to explain alternations in terms of an underlying
structure assumed a priori to constrain the types of changes that are possible.
NOTES
1. In using the term “principles of chain shift” to refer to principles that may be invoked to model
uncoordinated changes in single vowels, I follow Labov (1994:117), who stated that, “though [Prin-
ciples I through III] are stated in terms of chain shifts, I will not hesitate to use them to describe and
classify individual movements where they apply.” “Chain-shift model” is thus used as a shorthand
term to refer to structuralist analyses of the type elaborated by Labov, whether or not any actual chain
shift is taking place.
2. The “evaluation problem” (Weinreich, Labov, & Herzog, 1968) continues to pose difficult
questions in sociophonology. Since as yet only a small-scale preliminary study of listener reactions to
the variants of Tyneside face and goat has been carried out (Holmes, 2000), and since the evidence
for social attitudes toward these variants is rather indirect and patchy, for present purposes it is
assumed that disparities in the distribution of variants across speaker groups can be taken as evidence
of differences in the variants’ evaluation.
3. That is, the use of [V] pre-vocalically (as in [bVEd] bread) and [f,v] for (T,D) (as in [fIn] thin or
['mUvá]mother).
4. While collapsing detailed transcriptions into categories such as Type I,Type II, Type III may be
undesirable in the sense that one loses the phonetic resolution of the original transcriptions, it should
beremembered thattheterm “variant”is only meaningfulif wechoose toimpose categories ontowhat
is, after all, a phonetic continuum. As long as this is carried out in a careful, principled, and reproduc-
ible way, the approach serves the sociolinguist’s purposes well; the advantages of very detailed im-
pressionisticor instrumental approachesareusually outweighed bythedifficultiesthey presentin terms
of handling the amounts of data needed for an adequate overview of the variation within the commu-
nity.For assessmentsofthe reliability ofphonetictranscription, see Cucchiarini(1996),Nairn and Hur-
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 97

ford(1995), Vieregge (1987);on problemswith thetraditionalformant-frequency analysisof vowels,
see Faber and di Paolo (1995), Harrington and Cassidy (1994), Pisoni (1997).
5. The choices available were (a) northeast, (b) northwest, (c) east midlands, (d) west midlands,
(e) southeast, (f) southwest, (g) Standard English, (h) north Wales, (i) northern Scotland, (j) Belfast
area, (k) no response.
6. The membership of the goat lexical set in older TE is particularly difficult to fix. Both, for
instance, is transcribed by Jones as [bjET], while hold is [ha:d] and so is [si:]. Such pronunciations
continue to be used in modern conservative TE, but they are relatively rare and alternate with the
goat pronunciations described in more recent accounts. Curiously, Jones transcribed items that are
members of the strut set in accents featuring a foot ;strut distinction, such as come,done,fun,
but,numb,mother, with [o], but showed good (the sole foot item) as [gud]. It is doubtful, however,
whether Jones meant to imply a foot ;strut split here.
7. See also Trudgill (1998).
8. The results of a small-scale pilot study (8 speakers, or one per cell) also confirm the consistency
of the transcriptions. For variants of Types I, II, and III the correlations between face and goat
achieved a high level of statistical significance.
9. Implemented using NAG Generalized Linear Models (GLIM), v. 3.77.
10. A paucity of face tokens in Vb,Vz,and Vl contexts led to the omission of these contexts in the
analysis. The context effects reported here are therefore based upon the V#,Vp,Vs,and Vn contexts
only. Interestingly, Labov,Yaeger, and Steiner (1972:104) stated that following nasals may promote
the raising of tense ingliding vowels, an observation that is backed up to some extent in the distribu-
tions of [I@] and [U@] among these TE speakers and that matches results reported by Kerswill (1984)
for the face vowel in Durham, a city of some 82,000 inhabitants around 15 miles (25 km) to the south
of Tyneside.
11. A reviewer suggested that Type III diphthongs may appear more often in WL style as a conse-
quence of a reduction in speech rate. Specifically, the offglide, which is perhaps the result of coartic-
ulation of the vowel with the following consonant, could be more perceptible at slower rates.
12. This effect is weakened when the localized variants are considered individually because of the
complexinteraction betweenthe trendsfor[U@] and[Õ:] broughtaboutby youngerMC men’satypical
preference for the latter variant.
13. Labov’s transcription system corresponds to Wells’s lexical set keywords as follows: 0ow05
goat,0uw05goose; the 0ay0and 0aw0vowels mentioned in connection with the Martha’s Vineyard
study are price and mouth, respectively.
14. At least in descriptions of adult speech; phonological analyses of child speech tend to attach
more importance to implicational dependencies obtaining between consonants in language acquisi-
tion (see, e.g., Menn & Stoel-Gammon, 1995).
15. Fitting symbols representing the set of phonological contrasts into a language-universal grid or
matrixis ofcourse visuallyuseful wherephonologies ofindividuallanguages areto becompared, but,
as Simpson (1999) pointed out, this sort of generalization by necessity involves a good deal of ide-
alization and is undertaken at the expense of preserving much relevant phonetic detail.
REFERENCES
Auer, Peter, & Hinskens, Frans. (1996).The convergence and divergence of dialects in Europe: New
and not so new developments in an old area. Sociolinguistica 10:1–30.
Beal, Joan C. (1985). Lengthening of ain Tyneside English. In R. Eaton, O. Fischer, W. Koopman, &
F. van der Leek (eds.), Papers from the 4th International Conference on English Historical Lin-
guistics, Amsterdam. Amsterdam: John Benjamins. 31–44.
_
(1993). The grammar of Tyneside and Northumbrian English. In L. Milroy & J. Milroy (eds.),
Real English: The grammar of English dialects in the British Isles. London: Longman. 187–213.
Cheshire, Jenny, Gillett,Ann, Kerswill, Paul, & Williams, Ann. (1999). The role of adolescents in di-
alect levelling. Final report submitted to the UK Economic & Social Research Council, June 1999.
Crothers, John. (1978). Typology and universals of vowel systems. In J. H. Greenberg (ed.), Univer-
sals of human language. Vol. II: Phonology. Stanford, CA: Stanford University Press. 93–152.
Cucchiarini, Catia. (1996). Assessing transcription agreement: Methodological aspects. Clinical Lin-
guistics and Phonetics 10:131–155.
Disner, Sandra F. (1980). Insights on vowel spacing: Results of a language survey. UCLA Working
Papers in Phonetics 50:70–92.
98 DOMINIC J.L. WATT

Docherty, Gerard J., & Foulkes, Paul. (1999). Derby and Newcastle: Instrumental phonetics and
variationist studies. In P. Foulkes & G. J. Docherty (eds.), Urban voices: Accent studies in the
British Isles. London: Arnold. 47–71.
Docherty, Gerard J., Foulkes, Paul, Milroy, James, Milroy, Lesley, & Walshaw, David. (1997). De-
scriptive adequacy in phonology: Avariationist perspective. Journal of Linguistics 33:275–310.
Faber,Alice, & di Paolo, Marianna. (1995). The discriminability of nearly merged sounds. Language
Variation and Change 7:35–78.
Foulkes, Paul, & Docherty, Gerard J. (eds.). (1999). Urban voices: Accent studies in the British Isles.
London: Arnold.
Harrington, Jonathan, & Cassidy, Stephen. (1994). Dynamic and target theories of vowel classifica-
tion: Evidence from monophthongs and diphthongs in Australian English. Language and Speech
37:357–373.
Hinskens, Frans. (1998). Dialect levelling—A two dimensional process. Folia Linguistica 32:35–51.
Hockett, Charles F. (1958). A course in modern linguistics. New York: Macmillan.
Holmes, Helen. (2000). Phonetic variation and percepts of accentedness. Unpublished BSc disser-
tation, University of Newcastle upon Tyne.
Hughes, Arthur, & Trudgill, Peter. (1979). Englishaccents anddialects: An introductionto socialand
regional varieties of British English. London: Arnold.
Jones, Daniel. (1911). English: Tyneside dialect (Northumberland). Le maître phonétique 26:184.
Kerswill, Paul. (1984). Social and linguistic aspects of Durham (e:). Journal of the International
Phonetic Association 14:13–34.
Krause, Christine B. (1989). A computerized analysis of the stressed vowel systems of Lowick0
Northumberland. Masters dissertation, Ruprechts-Karls-Universität, Heidelberg.
Labov, William. (1963). The social motivation of a sound change. Word 19:273–309.
_
(1991). The three dialects of English. In P. Ecker t (ed.), Newways of analyzingsound change.
Orlando, FL: Academic. 1–44.
_
(1994). Principles of linguistic change. Vol. I: Internal factors. Oxford: Blackwell.
Labov, William, Yaeger, Malcah, & Steiner, Richard. (1972). A quantitative study of sound change in
progress. Philadelphia: US Regional Survey.
Lass, Roger, (1989). System-shape and the eternal return: Front rounded vowels in English. Folia
Historica Linguistica 10:163–198.
_
(1990). A‘standard’ South African vowel system. In S. Ramsaran (ed.), Studies in the pro-
nunciation of English: A commemorative volume in honour of A.C. Gimson. London: Routledge.
272–285.
Lennig, Matthew. (1978). Acoustic measurement of linguistic change: The modern Paris vowel sys-
tem. Doctoral dissertation, University of Pennsylvania.
Liljencrants, Johan, & Lindblom, Björn. (1972). Numerical simulation of vowel quality systems: The
role of perceptual contrast. Language 48:839–862.
Luthin, Herbert W. (1987). The story of California (ow): The coming-of-age of English in California.
In K. M. Denning, S. Inkelas, F. C. McNair-Knox, & J. R. Rickford (eds.), Variation in language:
NWAV-XV at Stanford. Stanford, CA: Stanford University Department of Linguistics. 312–324.
Menn,Lise, &Stoel-Gammon, Carol.(1995). Phonological development.InP.Fletcher& B.MacWhin-
ney (eds.), The handbook of child language. Oxford: Blackwell. 335–359.
Mess, Henry A. (1928). Industrial Tyneside: A social survey made for the Bureau of Social Research
for Tyneside. London: Ernest Benn.
Milroy, James. (1995). Investigating the Scottish Vowel Length Rule in a Northumbrian dialect.
Newcastle and Durham Working Papers in Linguistics 3:187–196.
_
(1996).Acurrent change inBritish English:Variationin (th)inDerby.Newcastle andDurham
Working Papers in Linguistics 4:213–222.
Milroy, Lesley, Milroy, James, Docherty, Gerard J., Foulkes, Paul, & Walshaw, David. (1999). Pho-
nological variation and change in contemporary English: Evidence from Newcastle upon Tyne and
Derby. In J. C. Conde-Silvestre & J. M. Hernández-Campoy (eds.), Variation and linguistic change
in English: Diachronic and synchronic studies. Murcia: Secretariado de Publicaciones e Intercam-
bio Científico. 35–46.
Nairn, Moray J., & Hurford, James R. (1995). The effect of context on the transcription of vowel
quality. In J. Windsor Lewis (ed.), Studies in general and English phonetics in honour of Professor
J. D. O’Connor. London: Routledge. 96–120.
O’Connor, J. D. (1947). The phonetic system of a dialect of Newcastle-upon-Tyne. Le maître phoné-
tique 87:6.
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 99

Orton, Harold, & Halliday, Wilfrid J. (1963). Survey of English dialects (B): The basic material. Vol.
I: The six northern counties and the Isle of Man. Leeds: Arnold.
Orton, Harold, & Barry, Michael V. (1969–71). Survey of English dialects (B): The basic material.
Vol. II: The west midland counties. Leeds: Arnold.
Pfalz,Anton.(1918). ReihenschritteimVokalismus. Beiträgezur Kundederbayerisch-österreichischen
Mundarten,ch. 2.Vienna:Akademie derWissenschaften inWien, Philosophisch-historische Klasse,
Sitzungsberichte 19002.
Pisoni, David B. (1997). Some thoughts on “normalization” in speech perception. In K. Johnson &
J. W. Mullennix (eds.), Talker variability in speech processing. San Diego, CA: Academic. 9–32.
Rietveld, Toni, & van Hout, Roeland. (1993). Statistical techniques for the study of language and
language behaviour. Berlin: Mouton de Gruyter.
Rydland, Kurt. (1995). The Orton corpus and Northumbrian phonology: The material from Bam-
burgh and Bellingham (Northumberland). English Studies 76:547–586.
Samuels, Michael L. (1972). Linguistic evolution, with special reference to English. Cambridge:
Cambridge University Press.
Schwartz, Jean-Luc, Boë, Louis-Jean, Vallée, Nathalie, &Abry, Christian. (1997). Major trends in
vowel system inventories. Journal of Phonetics 25:233–253.
Simpson,AdrianP.(1999). Fundamentalproblems incomparativephonetics and phonology:DoesUP-
SID help to solve them? In J. J. Ohala et al. (eds.), Proceedings of the 14th International Congress
ofPhoneticSciences, SanFrancisco, August1999. Berkeley:University ofCalifornia Press.349–352.
Stewart, Ann H. (1976). Graphic representation of models in linguistic theory. Bloomington: Indiana
University Press.
Thelander, Mats. (1982). Aqualitative approach to the quantitative data of speech variation. In S.
Romaine (ed.), Sociolinguistic variation in speech communities. London: Arnold. 65–83.
Trubetzkoy, Nikolai S. (1969). Principles of phonology (translated by Christiane A. M. Baltaxe).
Berkeley: University of California Press.
Trudgill, Peter. (1998). The great East Anglian merger mystery (Cahiers de l’Institut de Linguistique
et des Sciences du Langage 11). Lausanne: Université de Lausanne. 415–424.
Vallée, Nathalie, Schwartz, Jean-Luc, & Escudier, Pierre. (1999). Phase spaces of vowel systems: A
typology in the light of the Dispersion-Focalization Theory (DFT). In J. J. Ohala et al. (eds.),
Proceedings of the 14th International Congress of Phonetic Sciences, San Francisco, August 1999.
Berkeley: University of California Press. 333–335.
Veatch, Thomas C. (1991). English vowels: Their surface phonology and phonetic implementation in
vernacular dialects. Doctoral dissertation, University of Pennsylvania.
Viereck, Wolfgang. (1966). Phonematische Analyse des Dialekts von Gateshead-upon-Tyne0Co.
Durham. Hamburg: Cram, de Gruyter & Co.
_
(1968). Adiachronic-structural analysis of a northern English urban dialect. Leeds Studies in
English 2:65–79.
Vieregge, Wilhelm H. (1987). Basic aspects of phonetic segmental transcription. In A. Almeida &A.
Braun (eds.), Probleme der phonetischen Transkription (Zeitschrift für Dialektologie und Linguis-
tik Beihefte 54). Stuttgart: Franz Steiner Verlag. 5–48.
Watson, Catherine I., Harrington, Jonathan, & Evans, Zoë. (1998). An acoustic comparison
between New Zealand and Australian English vowels. Australian Journal of Linguistics 18:185–
207.
Watt, Dominic. (1999). Phonetic variation in two Tyneside vowels: Levelling in lockstep. In J. J.
Ohala et al. (eds.), Proceedings of the 14th International Congress of Phonetic Sciences, San Fran-
cisco, August 1999. Berkeley: University of California Press. 1621–1624.
Watt,Dominic,& Milroy, Lesley.(1999). Patternsof variationand change inthree Newcastlevowels:
Is this dialect levelling? In P. Foulkes & G. J. Docherty (eds), Urban voices: Accent studies in the
British Isles. London: Arnold. 25– 46.
Watt, Dominic, &Tillotson, Jennifer. (1999). Aspectrographic study of vowel fronting in Bradford
English. Leeds Working Papers in Linguistics and Phonetics 7:206–234.
Weinreich, Uriel, Labov, William, & Herzog, Marvin. (1968). Empirical foundations for a theory of
language change. In W. Lehmann & Y. Malkiel (eds.), Directions for historical linguistics. Austin:
University of Texas Press. 95–189.
Wells, John C. (1982). Accents of English (3 vols.). Cambridge: Cambridge University Press.
Williams,Ann, & Kerswill, Paul. (1999). Dialect levelling: Change and continuity in Milton Keynes,
Reading and Hull. In P. Foulkes, & G. J. Docherty (eds.), Urban voices: Accent studies in the
British Isles. London: Arnold. 141–162.
100 DOMINIC J.L. WATT

APPENDIX
WORD LIST
sheet boat lap it half-cut
beetle total apron automatic
metre motor matron Jupiter
I beat it I wrote it micro epileptic
gate put metro sheet
paint footer leprosy read
fatal put it in petrol breeze
later boot acrid key
I hate it Bootle atlas gate
eighty-eight hooter hopper made
bet bite butter may
bent title hacker boat
felt mitre topple load
fettle pint bottle go
better bite it hackle boot
I met him out whisper brood
hat fount custard booze
ant outer after brew
battle pit whisker out
batter bitter doctor loud
drat it brittle chapter cow
cart print jumper sight
can’t I hit it hunter side
carter hilt bunker size
pot beak appear sigh
totter wreck attend sighed
bottle back occur knife
font I seek it appearance five
salt I wreck it attendance knives
I got it I back it occurrence dive
caught bank alpine dial
daughter lamp alter Friday
chortle leap polka diary
haunt cap staircase
I bought it steep it half-past
I’ve got to do it tomorrow
I had to put it off
He meant what he said
He’s booking separate tables for supper
A simple sentence
Pick up a packet of firelighters
Pack it in or beat it
He’s putting it off
He put in a bid
Jump up on the tractor
He won’t do that in a hurry
Put a comma in it
PHONETIC PARALLELS BETWEEN TYNESIDE VOWELS 101