This study (1) investigates social variation in the speech of Coleraine, an Ulster-Scots town in east county Londonderry, sixty miles north of Belfast. It is based on thirteen hours of recordings made for doctoral studies at the University of Ulster in 1981.(2) The author carried out field work using a family-based model that involved interviewing three generations (i.e. teenagers/young adults, parents, and grandparents) of four families, totalling 26 speakers. She followed Lesley Milroy's 'friend of a friend' approach to gain access to social and family networks and then correlated results with four non-linguistic variables: age, sex, areal differences (rural vs. urban), and social class (as shown by the speaker's type of housing — council-owned vs. private).(3) Speakers were recorded in peer-group interaction for 60 minutes and were then asked to read a word-list. Before turning to details of the variation itself, the author presents an overview of distinctive features of Ulster-Scots phonology.
The Ulster-Scots vowel system is not phonemically dichotomous (i.e. long vs. short), but rather exhibits a three-way classification into long vowels, short vowels and vowels that follow the Scottish Vowel Length Rule (SVLR).(4) SVLR vowels /i, u, o/ vary in length according to context, and length differences may be accompanied by noticeable phonetic differences. Figure 1 shows articulatory placement and vowel classification.
KIT [ɪ] (Wells 2.2.1)(5) This vowel is noticeably lowered and centred compared to Received Pronunciation (RP) and tends to be fronted before velars and palatals and to be further back and centralised before sonorants and laterals. In local vernacular it is lowered and centralised to approximate [ə] and may amalgamate with [ʌ], especially in male speech. Neutralisation is avoided in some male and elderly speakers, since [ʌ] is in turn lowered, backed and rounded to approximate [ɒ] or [ʊ].)
STRUT [ʌ] (Wells 2.2.5) This vowel may have lip rounding to approximate [ə] or [ɒ] in older working-class speakers. Its distribution excludes certain lexical sets of -OVE and -ONE words.
NEAR [i] (Wells 2.2.6) In broad vernacular speakers, this vowel may lower and back in polysyllabic short environments to become [ɪ], e.g. cheaper [tʃɪpr̩],
GOAT [o] (Wells 2.2.14) In long environments, this vowel may have closing [ʊ] or [ə] glide. Teenage middle-class females sometimes use [əʊ], following an RP model.
GOOSE [ʉ] (Wells 2.2.15) This close, rounded front or central vowel corresponds in distribution to /ʊ/ and /u/ phonemes in RP, producing such homonyms as pull = pool. Its height and frontness depend upon the phonetic environment, which also determines length. Long environments -z and -# result in highest variant [ʉ:]. Shorter environments produce lowered, retracted variants ranging progressively from [ʉ] before nasals and laterals, e.g. soon [sun] and pool [pʉl], to [ʊ] in shortest environment (i.e. before voiceless plosives), e.g. soup [sʊp], boot [bʊt].
All long monophthongs tend to have a closing centralised glide.
FACE [e] (Wells 2.2.11) This vowel approximates Cardinal Vowel 2, but is subject to the influence of phonological and sociolinguistic variation. It may be lowered and lengthened to [ɛ:] before velar plosives to neutralise with the /ɛ/ phoneme, e.g. bake = beck [bɛːk]. It may be raised and backed to [ë] in short environments, e.g. place [plës] or even [plɪs], and in disyllabic and polysyllabic words in older vernacular, e.g. table [tëbl̩], available [əvëləbl̩], Elderly speakers tend to use monophthong [eː], a rural, recessive variant noted in general use forty years ago by Robert Gregg.(6) Middle-aged and younger speakers tend to use [ɪə] or [eə] (a Belfast variant coming into Ulster-Scots speech)(7) in all environments except open syllables, where it remains principally monophthongal. Note the morphemic distinction: days [dɛːz] or [deiz] versus daze [dɪəz]. 'Corrected middle-class' style amongst younger females approximates RP with a closing glide [eɪ]
DRESS [ɛ] (Wells 2.2.2) This vowel approximates Cardinal Vowel 3, but is more open and centralised and tends to have a closing schwa glide. In pre-velar or prepalatal position, it has closing [i] glide, e.g. leg [lɛig].
TRAP [a] (Wells 2.2.3) This vowel is open and slightly back of centre. It corresponds to the /æ/ and /ɑ/ phonemes of RP, e.g. Sam = psalm [saːm]. An additional lexical set with this vowel is heard mainly in middle-aged, elderly and rural speech, e.g. wash [waʃ], warn [warn], once [wanst]. A recessive variant /a/ + velar consonant neutralises with /ɛ/ in elderly working-class males, e.g. bag [bɛg], factory [fɛktrɪ].
NORTH [ɔ] (Wells 2.2.22) This vowel is half-open, back and rounded between Cardinal Vowel 5 and Cardinal Vowel 6. It corresponds to the /ɒ/ and /ɔ/ phonemes of RP, but also includes lexical sets of -OVE and -ONE words, e.g. gloves [glɔːvz], above [abɔːv], done [dɔːn].
PRICE [aɪ] (Wells 2.2.16) Note that there is a distinction between [aɪ] in long environments and certain lexical items (e.g. tide, rise, my, dye) and [ʌi] in short environments (e.g. eye, die, white). All ages and particularly males have a strong tendency in local vernacular to raise the first element to [ɛ] in all phonetic environments, e.g. nine [nɛɪn]. Note that there is less phonetic distinction between [aɪ] and [ʌi] than that observed in Ulster-Scots by Gregg, who noted [aeː] versus [əi].(8) This greater distinction is heard only in elderly informants. It is probable that with the raising of the first element in both cases to [ɛ] the phonemic distinction is being lost, bringing the accent in line with Belfast phonology(9) and RP phonemic distribution.
MOUTH [aʉ] (Wells 2.2.18) In long environments the first element of this diphthong is lengthened and in local vernacular speech it is raised to [ɛ], as in Belfast vernacular.(10)
/l/ is subject to considerable sociolinguistic variation. Elderly speakers use little or no dark /l/, reflecting the older Ulster-Scots clear /l/ speech pattern as described by Gregg.(11) In contrast, teenage and young adults use a high percentage of dark /l/, as shown in figure 2. The shift towards dark /l/ is not initiated by a desire to approximate standard British English since these younger speakers show equally high dark /l/ scores in the non-standard context of syllable-medial position. Rural-born and middle-class speakers use more clear /l/, and there is a shift led by working-class males away from rural, clear /l/ towards urban, dark /l/, both varieties being nonstandard. There is strong evidence to indicate that dark /l/ is a marker of male urban working-class vernacular and is increasing in all phonological environments, ages and sexes despite certain middle-class pressure to maintain clear /l/ in syllable-final position.(12)
/ŋ/ This phoneme is always realised by the variant [n] by working-class speakers in the local vernacular.
/r/ [ɹ] This consonant is characteristically rhotic, with an open, frictionless continuant similar to North American /r/. The vocalic system in environments before /r/ is linguistically unstable, with several subsystems undergoing change and merger, presenting an intermediate stage of urbanisation between Ulster-Scots and Belfast systems.
/ð/ is frequently deleted in intervocalic position in male working-class speakers, e.g. mother [mɔər], Final /ð/ may be devoiced, e.g. with [wɪθ].
/s/ is frequently retracted in all environments in middle-aged and elderly speakers.
/ʍ/ occurs in most speakers (less in the younger generation) in initial position in wh- words, e.g. what [ʍɔt],
/x/ is used by all speakers, mainly in place and family names, dialect words, and exclamations, e.g. Donaghadee [dɔnəxədi], Donaghy [dɔnəxɪ], dreigh [d̪rix], ugh [ʉx].
In male urban working-class vernacular, all voiceless, intervocalic plosives tend to be voiced, e.g. get it [gɛdɪʔ], happy [habɪ], pack of [pag əv]. Yet unstressed syllables ending in [ɪd] may be devoiced, e.g. stupid [stjʉpɪt], ended [ɛndɪt]. The glottalic air stream mechanism is frequently used for final voiceless plosives when followed by a pause, e.g. cat [kaʔtˈ]. Alveolar plosives are dentalised when immediately followed by /r/ in some middle-aged and elderly speakers, e.g. Antrim [ant̪rim]. Velar plosives are articulated further forward in the mouth than their RP counterparts, giving a palatalised quality and influencing adjacent sounds, e.g. bag [baig]. Similarly all affricates and palato-alveolar fricatives tend to be palatalised with lack of lip rounding, causing a raising of preceding vowels, e.g. mash [maiʃ], patch [paitʃ].
Intervocalic /t/ is characterised by considerable sociolinguistic variation. In particular, [tʔ] tends to be used in 'corrected' rural speech; [ʔ] is used by all informants, but most often by females and rural speakers; and [d] occurs most often in male, working-class urban vernacular. [ø] is an advanced male working-class variant.
This study examines the sex-based rivalry between the two predominant vernacular variants of intervocalic /t/ in Coleraine speech (i.e. the glottal stop [ʔ] and [d]). The family-based model was particularly successful in gaining access to female networks, and results of the study challenge some widely held assumptions about female speech — namely that women stereotypically use more standardised variants in contrast to men, who use more non-standard variants. While there is substantial evidence to support the conventional theory, the author would agree with Fasold, who notes that 'the consistency of the gender pattern has sometimes been overstated'.(13) Some results of this study support the reverse trend, but the variation probably depends on which social networks are chosen for study. For example, this study focussed on working-class families, which had close-knit, female social networks centred in and around the home and local church, in contrast with the male network structure, which was more open and loose-knit. It was therefore not surprising that female scores for some non-standard variants (such as [n] for /ŋ/) were higher than male scores.
Glottal stop appears to be undergoing rapid changes in both standard and non-standard varieties of British English. It has been a popular variable in recent studies due to its widespread use and negative prestige. However, it must be stressed that there are some seriously misleading categorisations of glottal behaviour in many studies, due to inadequate description and also insufficient consideration of standard versus non-standard usage in British English. It was therefore essential for this study to create more precise articulatory definitions and also make a detailed distinction between standard and non-standard usage.(14) Historically, glottal stop has been described as a typical feature of Ulster-Scots, whereas there is no record of the intervocalic [d] variant in Ulster-Scots,(15) although it commonly occurs in Belfast speech,(16) Mid Ulster speech generally, and also West Country speech in England.(17)
In this study three separate phonological environments have been examined for intervocalic /t/: 1) word-initial, as in go to (where the word to functions enclitically); 2) word-medial, as in water, and 3) word-final, as in got a. There were two main reasons why context was so crucial. Word-initial environments occurred at only 15% while the other two environments were equally matched at 42%. Total scores would have hidden this bias. Also, some variants were more context-dependent than others. For example, [d] occurred more in word-initial context, which is least frequent, whereas glottal stop occurred most in the more frequent word-medial position. Thus, it was crucial to isolate these contexts; otherwise subtle trends would have been masked or even obliterated.
Results show that in all environments and ages, males use more [d], whereas females use more glottal stop. In fact, the two variants are almost in complementary distribution in initial and final environments, with the most extreme scores in teenagers. Other age groups show similar trends to a lessening degree with age. We might conclude that [d] is a marker of male style and that glottal stop is a marker of female style, as shown in final environment in figure 4 where scores are contrastively matched. However, this would be an oversimplification, since in the other two environments, sex-based linguistic behaviour is more complex. According to Labov et al., phonological conditioning does not occur in stable phonemes, and therefore these results would seem to indicate sound change in progress.(18) There is strong rivalry between the two variants that is not only sex- and age-based, but also phonologically conditioned. But which is the incoming form, and which is the receding form, if there is a receding form? It has already been noted that in earlier Ulster-Scots studies glottal stop was a familiar feature, whereas the intervocalic [d] variant was not mentioned. Therefore one could hypothesise that [d] is an incoming variant from influential larger, urban centres such as Belfast, its progress being led by males and younger females, as is typical of vernacular changes. However, glottal stop is also increasing in Coleraine speech. Figure 3 graphically shows that in all age groups and both sexes it is increasing most steadily in word-medial position with women strongly in the vanguard of this change. We may conclude that the three environments, as illustrated in figures 3, 4, and 5, highlight an interesting sex-based conflict between two non-standard variants, both of which appear to be increasing as they move through the various barriers of phonological environment, sex and age of speaker.
If this hypothesis is correct, then word-medial environment in figure 3 represents an earlier stage of the ongoing change. The predominant variant here for all speakers is glottal stop, which is increasing in younger speakers. The [d] variant shows a much lower incidence and is strongly associated with male speech. It presents Labov's classic pattern of a stable linguistic marker, since it does not appear to be participating in any change.(19) Note that females tend to avoid [d] almost entirely in this environment.
Figure 4 (i.e. word-final environment) shows a striking pattern of sex-based complementary distribution of glottal stop and [d] variant, with the cross-over pattern indicating change in progress.(20) Clearly there is rivalry between the variants, and all speakers show an increased use of [d] variant with males in the vanguard. Sex differences are strikingly obvious, with males using more [d] and females using more glottal stop in all age groups.
The trend towards [d] is at its most advanced in word-initial environments, and figure 5 shows the steady increase in [d], led by men and teenage girls, with a corresponding decrease in glottal stop. If the three environments are collapsed together, as in figure 6, we find a challenge to the hypothesis that [d] is beginning to increase at the expense of glottal stop. Younger speakers of both sexes show a strong preference for glottal stop, with younger females leading the way. As predicted, the less frequent word-initial environment of figure 5, which showed [d] to be on the increase, is totally submerged beneath the larger glottal stop scores of the other two environments.
To summarise, there appear to be two sex-based trends operating in Coleraine vernacular, both of which are moving away from standard British speech, [d] is the incoming, male-led variant from Belfast, but glottal stop is also increasing and is strongly preferred by females. Fasold suggests that sex may be a more influential factor on language than social class, i.e. gender differences may typically precede social-class differences, which occur at a later stage.(21) He says that rather than men choosing lower-status variants and women choosing higher-status ones (as has been generally thought), perhaps 'there are ways of speaking which men use to emphasize their masculinity and other forms women use to symbolize their femininity, and that this is more basic than social class'.(22) The results of this study tend to support these views, with the [d] variant being used almost exclusively by men, and the glottal variant preferred by women. Fasold goes on to suggest that 'perhaps men's speech became derivatively associated with lower social status and women's speech with higher status'.(23) In Coleraine vernacular there is evidence of some social ranking of non-standard variants, in that female variants appear to have higher social prestige. For example, in figures 7 and 8, which correlate glottal stop and [d] variant with housing and sex, we find a clear correlation between housing type and choice of variant, which is further complicated by sex contrasts.
Figure 7 shows a striking housing-based and gender-based pattern of complementary distribution for glottal stop. It appears to be ranked according to housing type for both sexes, but in contrasting patterns, with working-class females in government housing and middle-class males in private housing using most glottal stop. In comparison with the [d] data in figure 8, the [d] variant shows mainly a sex contrast. The strong class stratification in glottal stop probably shows that it is at a further stage of change than the [d] variant. Presumably if [d] becomes more accepted by female groups, it will also show more class stratification.
It is interesting to note that the preferred female nonstandard glottal form is also the older rural form. This may provide a clue as to some of the mechanisms of change. Bortoni-Ricardo's study of a Brazilian city found that as people move from close-knit, norm-enforcing rural networks to more loose-knit urban settings, they become more vulnerable to social (and linguistic) influences and change.(24) The urban/rural distinction is a highly significant factor in the present study of small town speech. As speakers move out of rural networks to a more urban setting (through, for example, the necessities of employment), they experience social pressures to conform to the new forms in order to avoid being perceived as old fashioned or rural and also in order to be accepted by the new urban network. According to James Milroy, 'avoidance of stigma (attached particularly to rural stereotypes) is a powerful initiator of rapid change. The stigmatised rural form may be replaced by an urban vernacular form ... which in turn may be stigmatised by upwardly mobile persons'.(25)
One vehicle for the innovation of /t/ voicing to [d] in Coleraine speech was probably via young male dock workers who had to travel back and forth to Belfast docks. As their identity with their home base weakened, they became more vulnerable to change and thus adopted and transported the urban [d] form back home. Thus [d] became associated with working-class male status. At the same time, women maintained their feminine distinctiveness by rejecting or resisting the newer male urban forms, preferring to use the more conservative, non-standard rural variant which may eventually achieve middle-class status (as has already happened in the case of clear /l/). It is noteworthy that in this study, working-class males are the innovators, whereas women tend to be more conservative.
The female preference for rural variants is significant in the light of previous research into attitudes towards varieties of British speech. Macaulay describes a predominant tendency for city speech to be overtly and strongly stigmatised whereas rural varieties are regarded as pleasant or attractive.(26) The Coleraine results not only confirm the well-documented covert male prestige of urban speech, but also the female-associated overt prestige of rural speech. It would appear that notions of status and stigma are to some extent sex-based. Coleraine working-class males are motivated by an avoidance of rural stigma (through peer pressures in the workplace), whereas the women in the sample are more motivated by conservative 'homestead' values of family, church and neighbourhood, together with avoidance of city stigma. This lends interesting support to the findings of Tannen, who studied differences between male and female world views as reflected in language.(27) She found that in general men are more focussed on hierarchies of power and accomplishment (involving independence, status and contest), whereas women are more focussed on hierarchies of friendship (involving intimacy, connection and community).
We have already seen that glottal stop is socially ranked, indicating its status and stigma. In general the Coleraine results show that, although both sexes displayed this ranking pattern to different degrees for different variables, it appeared most consistently in the scores of women. This leads to the hypothesis that in the process of change, non-standard variants become associated with status/prestige mainly via women. It may be that women are more sensitive to these social parameters and that social connotations are (consciously or unconsciously) applied or even initiated by women. James Milroy also suggests that women, who typically have a 'normalising influence', are involved in later stages of linguistic change, e.g. the rapid middle stage of lexical diffusion.(28) However, an alternative explanation could be that the Coleraine findings are simply the result of methodological bias, and a study by a male researcher focusing on male network structure may reveal similar social patterning in men.
By way of conclusion, the sex-based rivalry between the two nonstandard /t/ variants in Coleraine speech raises several questions. Will the two variants continue to strengthen in opposition to each others? Or will one become more associated with middle-class values (via females)? Glottal stop appears to be gaining increasing prestige in many urban varieties of British English.(29) On the other hand, will [d] continue to spread via the male chain, or will it become stigmatised by upwardly mobile speakers — as has been observed in Belfast.(30) Clearly more research is needed on these questions.
It must be stressed that the above discussion is limited to the Coleraine data only. Any conclusions may not apply to other small towns in which Ulster-Scots is spoken or indeed to other social networks in Coleraine.
(1) An earlier version of this paper was published in Belfast Working Papers in Language and Linguistics 11 (1996), 223-237.
(2) See Kingsmore, Rona K., Coleraine Speech - Phonology and Sociolinguistics (University of Ulster Ph.D. thesis, Coleraine, 1983); and Kingsmore, R. K., Ulster Scots Speech (Tuscaloosa, 1995).
(3) Milroy, Lesley, Language and Social Networks (Oxford, 1987).
(4) For further discussion of the Scottish Vowel Length Rule, see Aitken, A. J., 'The Scottish Vowel Length Rule' in Benskin, Michael, and Michael L. Samuels (eds.), So Mony People, Longages and Tonges: Philological Essays in Scots and Mediaeval English Presented to Angus Mcintosh (Edinburgh, 1980), 131-157; and Kingsmore, op. cit. (1983).
(5) Vowels are keyed to their 'standard lexical sets', as presented in Wells, J. C., Accents of English, 3 volumes (Cambridge, 1982).
(6) Gregg, R. J., 'Notes on the Phonology of a County Antrim Scotch-Irish Dialect: Part 1', Orbis 7 (1958), 392-406.
(7) Milroy, James, 'Synopsis of Belfast Vowels', Belfast Working Papers in Language and Linguistics 1 (1976), 111-116.
(8) Gregg, op. cit.
(9) Milroy, James, op. cit.
(11) Gregg, R. J., 'The Scotch-Irish Dialect Boundaries in Ulster' in Wakelin, Martyn R (ed.), Patterns in the Folk Speech of the British Isles (London, 1972), 110.
(12) For greater details of phonological and social conditioning, see Kingsmore, op. cit. (1995).
(13) Fasold, Ralph, Sociolinguistics of Language (Cambridge, 1992), 92.
(14) For further details, see Kingsmore, op. cit. (1995).
(15) Gregg, R. J., 'Scotch-Irish Speech in Ulster' in Adams, G. B. (ed.), Ulster Dialects: An Introductory Symposium (Holywood, 1964), 163-192.
(16) Milroy, James, Speech Community and Language Variety in Belfast. Report to Social Science Research Council, 1977.
(17) Wells, op. cit. (1982), vol. 1, 344.
(18) Labov, William, Malcah Yaeger, and Roger Steiner, A Quantitative Study of Sound Change in Progress 2 (Philadelphia, 1972), 263.
(19) Labov et al., op. cit. 179.
(20) Milroy, James, and Lesley Milroy, 'Linguistic Change, Social Network and Speaker Innovation', Journal of Linguistics 21 (1985), 371; Labov, William, Sociolinguistic Patterns (Philadelphia, 1972).
(21) Fasold, op. cit.
(22) Ibid., 99.
(24) Bortoni-Ricardo, S. M, The Urbanization of Rural Dialect Speakers: A Sociolinguistic Study in Brazil (Cambridge, 1985).
(25) Milroy, James, op. cit. (1977), 12.
(26) Macaulay, R. K. S., Language, Social Class and Education: A Glasgow Study (Edinburgh, 1977).
(27) Tannen, Deborah, You Just Don't Understand (New York, 1990).
(28) Milroy, James, 'The Methodology of Urban Language Studies: The Example of Belfast' in Little, David (ed.), Perspectives on the English Language in Ireland. Proceedings of the First Symposium on Hiberno-English (Dublin, 1986), 42.
(29) Wells, op. cit., 323; Mees, Inge, 'Glottal Stop as a Prestigious Feature in Cardiff English', English World-Wide 8 (1987), 37.
(30) Milroy, James, op.cit. (1977), 12.