The phonetic and social correlates of non-rhoticity and derhoticised /r/ in Edinburgh English

The study of rhoticity has attracted a lot of interest in sociophonetic studies of urban Scottish speech. This work indicates a complex patterning in the realisation of postvocalic /r/ across the socioeconomic spectrum. The evidence points to rhotic approximants occurring in Middle Class speech, and derhoticisation (or a pharyngealized vowel; Speitel & Johnston 1983) typifying Working Class speech, both in Edinburgh (Romaine 1978; Lawson et al. 2008, et seq.; Scobbie et al. 2008, 2013; Schützler 2010, et seq.) and Glasgow (Macafee 1983; Stuart-Smith 1999, et seq.). Lawson et al. (2014: 63) model this Scottish /r/ variation along a 7-step phonetic continuum from deletion to derhoticisation to several approximant types and lastly to taps and trills. Recent research (Dickson & Hall-Lew 2015) has demonstrated that upwardly mobile speakers, who are intermediate in the socioeconomic scale, show the highest rate of approximant use, which suggests that rhoticity has a non-linear relationship with class. Furthermore, although acoustically similar, the non-rhotic and derhotic variants are maximally distinct in social distribution, used respectively by Middle Class women and Working Class men. Despite this, auditorily distinguishing non-rhotic and derhotic variants is notoriously difficult, such that even phonetically trained Edinburgh natives can struggle with coding them reliably (cf. Stuart-Smith et al. 2014). Thus, while Lawson et al. have evidenced a clear articulatory difference between the non-rhotic and derhotic forms, the question remains as to whether the variants do constitute auditorily distinct realisations of /r/, or whether instead the indexicality of a post-vocalic /r/ is signalled entirely by the quality of the preceding vowel. We pursue this question by examining two of the several acoustic measures of derhotic /r/ described by Stuart-Smith et al. (2014) in a corpus of spontaneous speech from 16 male and female Edinburgh speakers aged 57-69, recorded in 2013. The sample represents three broad socioeconomic groups: the Working Class (WC), the upwardly mobile New Middle Class (NMC), and the Established Middle Class (EMC). Our first analysis consists of identifying instances of the ‘breathy period’ (Lawson et al. 2008) or ‘audible frication’ (Stuart-Smith et al. 2014) that is often found to occur for derhotic variants at the vowel offset, coding for its presence/absence and duration. In our second analysis we measured the F1 and F2 of the approximate midpoint of the preceding vowel for the subset of tokens belonging to the start lexical set. We find that in 135 utterance-final tokens from 13 speakers, 51% of the data were realised with some frication. 87% of the fricative /r/s were produced by Working Class speakers and 90% were produced by men. The data also indicate that, overall, the duration of frication is longest for the WC speakers. While the presence or absence and duration of this breathy period seem to be a promising cue for distinguishing the variants, a number of complications arose with this analysis. Namely, the derhotic breathy period is really only discernible for utterance-final tokens and, among those, the duration measurements are highly variable, partly due to differences in amplitude quality as captured by the field recorder. The results for the start vowel show a strong interaction between gender and social class. The EMC women favour a higher, backer realisation of /ɑ/ (consistent with non-rhotic Anglo-English speech) compared to WC women, as was expected. The men on a whole produce a start vowel that is somewhere in between, and show little to no F1 or F2 differences according to socioeconomic class. Overall, we take these results to suggest that there are cases where it is possible to use acoustic measures to distinguish non-rhotic and derhotic variants, but in the (many) other cases where those measures are not feasible, the optimal definition of the /r/ variable considers the entire rime and takes the quality of the preceding vowel as a necessary parallel aspect of variation.