SECTION D EXTENSION: LINGUISTIC READINGS 192 EXTENSION: LINGUISTIC READINGS GLOTTALISATION IN CARDIFF In this extract from a larger study of the accent patterns of Cardiff in Wales, Beverley Collins and Inger Mees focus on glottalisation: the production or incorporation of a glottal stop when pronouncing the consonants /p/, /k/ and especially /t/. The Cardiff accent is unusual in British English in that glottalisation is regarded as a prestigious rather than stigmatised feature. Sociolinguistics and sociophoneticians are particularly drawn to anomalous phenomena such as this for the possible light that they shed on social patterns in general. The study excerpted below builds on earlier work over many years on the Cardiff accent. This allows Collins and Mees to make observations across history, comparing actual results from several studies in real-time. Another way of studying historical devel- opment is by investigating the speech across the age-ranges at any one chronological moment, which gives the researcher an apparent-time perspective. Collins and Mees correlate the accent patterns of their speakers with a complex of social variables to show the symbolic value of the accent feature. Beverley Collins and Inger Mees (reprinted from ‘Cardiff: a real-time study of glottali- sation’, in P. Foulkes and G. Docherty (eds) Urban Voices: Accent Studies in the British Isles (2000), London: Arnold, pp. 185–202) Historically, one can distinguish between three types of linguistic area in Wales: Type 1. The present-day Welsh heartlands where Welsh was overwhelmingly the majority language until 1900 (e.g. most of Gwynedd and Dyfed, together with parts of Clwyd and fragments of West Glamorgan and Powys). In these regions, this is either still the case or otherwise a large percentage of the population remains bilingual. Type 2. Areas which were largely Welsh-speaking until about 1850 but rapidly passed through stages of bilingualism towards English monolingualism in the early twentieth century, e.g. the industrialised former coal-mining valleys of south-east Glamorgan and south-west Gwent and border areas of Clwyd. Type 3. Border or coastal areas where English has been spoken by the vast majority of the population since well before 1800, in some cases possibly from the time of the Norman Conquest. This category includes not only marginal regions such as southern Pembrokeshire, south Gower, Powys and eastern Gwent, but also cru- cially the densely populated low-lying parts of southern Glamorgan and Gwent. (Collins and Mees 1991: 75–6; see Williams 1990 for a history of anglicisation in Wales, and Coupland and Thomas 1990: 8–9 for a slightly different typology of Welsh English) This [study] will deal with the core area of the last-named region, i.e. the city of Cardiff. [. . .] ‘Cardiff English’ (CE) [. . .] is sharply demarcated from the (mainly Type 2, periph- erally Type 1) speech of the rest of industrial South Wales, which we shall term General South Wales English (henceforth GSWE). [. . .] CE, unlike most British urban varieties, has comparatively little glottalisation. Furthermore, glottalisation decreases as one moves down the socio-economic scale, Beverley Collins and Inger Mees D1 GLOTTALISATION IN CARDIFF 193 so that the broadest accents are notable for the paucity of this feature (even though CE is still at variance with most other Welsh accents, especially those of Type 1, where glottalisation is hardly known; see Wells 1982: 388; Tench 1990: 131). Work carried out on Cardiff speech back in 1976 and 1981, based on a real-time study of the speech of schoolchildren, and reported in Mees (1983, 1987, 1990), showed that glottalisa- tion following a pattern of distribution found in RP (Wells 1982: 260–1) appeared to be on the increase in Cardiff. An analysis of the class, style, gender and time variation revealed glottalisation of word-final /t/ to be a spreading prestigious feature, with the process of change being led by young middle-class females (see Docherty, Foulkes, Milroy, Milroy and Walshaw (1997: 303–4) for comparable findings in respect of Tyneside). However, there was little evidence of the change having passed into the speech of the working classes, where glottalisation remained sporadic. The Cardiff research may be considered as a ‘panel study’. This has been defined by Labov (1994: 76) as a type of project that ‘attempts to locate the same individuals that were the subjects of the first study, and monitors any changes in their behaviour by submitting them to the same questionnaire, interview or experiment’. Labov continues: This is an expensive and time-consuming procedure, if it is planned as a panel study from the beginning, for the initial sample must be large enough to take the inevitable losses into account. An unplanned panel study will be left with a reduced sample, perhaps too small for statistical significance, but nonetheless extremely valuable for the interpretation of the original observations. Several linguists have returned to the community they originally studied and have replicated their work, but the pursuit of the same individuals was usually not a part of the basic design. [Labov 1994: 76] Fortunately, although the Cardiff project was indeed not originally envisaged as a panel study, it has nevertheless proved possible to track down an unusually high percent- age of the original informants. Labov (1994: 97) regards 50% as ‘a remarkably high proportion’; in Cardiff we managed to locate 67%. [. . . H]ere we intend to concen- trate on an in-depth scrutiny of a micro-group in order to investigate one aspect of the spread of glottalised variants. Mees (1983) reports the results of a comparison of data collected in the 1976 and 1981 fieldwork. The analysis dealt with the speech of 36 informants, who had been selected from the original sample and assigned to three social classes (Middle Middle Class, Lower Middle Class and Working Class, henceforth MMC, LMC, and WC) on the basis of father’s occupation – each class sample consisting of six boys and six girls. Two speech styles (interview style and reading passage style) were considered. An inves- tigation was undertaken of co-variation of six variables with social class, gender, style and time. The present discussion is concerned more particularly with the speech of WC female subjects and it is on this that we shall primarily focus – specifically on their realisations of word-final pre-consonantal and pre-pausal /t/ in a small set of seventeen (mostly monosyllabic) high-frequency words, i.e. it, bit, get, let, at, that, got, lot, not, what, put, but, might, right, quite, out, about. Unlike other words ending in /t/, these items appear to behave differently in CE with respect to realisational possibilities; cf. the Beverley Collins and Inger Mees 194 EXTENSION: LINGUISTIC READINGS discussion in Docherty et al. (1997: 297), and Milroy, Milroy and Hartley (1994: 24), for whether words of this kind should also be conjectured as forming a subset in Tyneside. In CE, most word-final /t/s can be produced as [t] or [m, mt], but for this set of com- mon words there is a third possible variant, namely elision (ø), whereby /t/ is either completely elided or else realised as a very weakly articulated stop with inaudible release. Tables D1.1 and D1.2 present the distribution of [t], [m, mt] and ø as realisations of pre-consonantal and pre-pausal /t/ for the females in three social classes at two points in time. From Table D1.1 it is clear that glottalised variants are favoured by the girls in the higher social classes at both points in time, whereas elided forms are characteristic of the working class. Pre-pausally, the picture is much the same, as illustrated by Table D1.2. Again, the high percentage of glottalised forms in the middle classes is striking. Conversely, the working-class girls clearly have a preference for the elided variant. Results from the fieldwork undertaken in 1990 allowed us to investigate the extent to which CE middle-class glottalisation has infiltrated into working-class speech. We selected from the original (1976) sample a micro-group of four working-class females for more intensive study, noting the occurrence of glottalisation in approximately 20 minutes of informal speech of each individual spread over the fifteen-year time span of the survey. We observed the pronunciation of word-final /t/ in three potential glottalisation sites: pre-consonantal, pre-pausal and pre-vocalic. Table D1.3 shows the percentage of glottalised as opposed to other variants of word-final /t/ in the interview style of the subjects concerned. Beverley Collins and Inger Mees Table D1.1 Percentage of [t], [v, vt] and ø for pre-consonantal /t/ (in interview style) for all females by social class at two points in time 1976 1981 [t] MMC 14.3 7.5 LMC 34.5 25.5 WC 37.4 35.1 [v, vt] MMC 79.2 84.6 LMC 30.8 47.6 WC 22.4 9.3 ø MMC 6.6 7.9 LMC 34.7 26.9 WC 40.2 55.7 N tokens analysed MMC 96 443 LMC 153 227 WC 198 281 GLOTTALISATION IN CARDIFF 195 Beverley Collins and Inger Mees Table D1.2 Percentage of [t], [v, vt] and ø for pre-pausal /t/ (in interview style) for all females by social class at two points in time 1976 1981 [t] MMC 33.9 8.5 LMC 46.4 20.6 WC 9.4 16.7 [v, vt] MMC 51.4 83.2 LMC 25.8 62.4 WC 21.2 15.7 ø MMC 14.7 8.3 LMC 27.8 17.1 WC 69.4 67.6 N tokens analysed MMC 49 264 LMC 76 157 WC 127 208 Table D1.3 Percentage of [t], [v, vt], [{] (i.e. voiced) and ø for word-final /t/ (interview style) for three contexts at three points in time: four subjects Word-final /t/ % [t] % [ V , V t] % ø % [ [ ] N tokens Pre-consonantal 1976 39.3 13.9 46.8 93 1981 31.7 7.4 61.0 165 1990 28.2 31.7 40.2 184 Pre-pausal 1976 27.4 23.8 48.8 53 1981 16.9 22.3 60.7 73 1990 26.7 44.7 28.6 72 Pre-vocalic 1976 21.9 2.3 10.3 65.5 60 1981 16.7 2.3 6.5 74.5 81 1990 30.0 11.7 8.3 50.0 131 . Cardiff accent is unusual in British English in that glottalisation is regarded as a prestigious rather than stigmatised feature. Sociolinguistics and sociophoneticians are particularly drawn to anomalous. Welsh heartlands where Welsh was overwhelmingly the majority language until 1900 (e.g. most of Gwynedd and Dyfed, together with parts of Clwyd and fragments of West Glamorgan and Powys). In. slightly different typology of Welsh English) This [study] will deal with the core area of the last-named region, i.e. the city of Cardiff. [. . .] ‘Cardiff English (CE) [. . .] is sharply demarcated