Sociophonology of rhoticity and r-sandhi in East Lancashire English
Barras, William Simon
Most discussions of English phonology argue that rhoticity and r-sandhi are necessarily in complementary distribution, citing the diachronic path that led to the loss of rhoticity and the resulting synchronic r ~ Ø alternations in non-rhotic dialects. However, some accounts suggest that ‘it would not be surprising to discover cases of intrusive-r in rhotic dialects’ (Harris 1994, see also Carr 1999, Uffmann 2007). In order to investigate how non-rhoticity and r-sandhi could be transmitted by dialect contact, this thesis uses data from speakers in five communities in Greater Manchester and East Lancashire. The locations are equally spaced along a twenty-mile route from Prestwich (a suburb of Manchester, where speakers are non-rhotic) to Accrington (a post-industrial mill-town which is on an ‘island of rhoticity’ (Britain 2009)). I show that individual speakers have variable levels of both rhoticity and r-sandhi, which matches research on early New Zealand English (Hay & Sudbury 2005). Beyond this key fact, I discuss several other aspects of the relationship between r-sandhi and rhoticity, including the phonological and dialectological significance of the patterns in the data. First, linking-r and intrusive-r have different distributions in my data, despite the typical claim that they are synchronically the same process. This supports the idea that speakers are sensitive to a difference between words with and without an etymological r: I attribute this to the influence of orthography and to sociolinguistic salience of intrusive-r. Second, the nature of my sample population allows me to consider both change in apparent time and variation across geographical space. An apparent time comparison of the distribution of non-rhoticity and intrusive-r in the five Lancashire locations shows that these features are spreading by wave diffusion: they reach nearby locations before they reach locations further away. However, there is also a pattern of urban hierarchical diffusion in which the most isolated and rural location, Rossendale, lags behind Accrington in its loss of rhoticity. This is examined in the light of local patterns of travel for work and leisure, which suggest that although Accrington is further than Rossendale from the non-rhotic ‘sea’ of surrounding speakers, socially constructed space is more significant than Cartesian distance in determining the amount of linguistic contact between speakers from different locations. Third, I show that levels of rhoticity are increasing for some young speakers in Rossendale, which supports the hypothesis that a local linguistic feature can have a ‘last gasp’ under pressure from a competing non-local feature before its eventual loss. However, the same speakers are also adopting intrusive-r more quickly than speakers from neighbouring areas and this is significant: while earlier research has suggested that the presence of hyperdialectal non-etymological r (e.g. lager [laôg@~]) can be an indication of a loss of rhoticity in progress, the East Lancashire data show a different situation, where non-etymological r is for the most part restricted to sandhi contexts. This shows that rather than r-colouring becoming part of the realisation of certain vowels (e.g. sauce [sO:ôs]), intrusive-r is becoming adopted as a hiatus-filling strategy: a phonological process is being used by some rhotic speakers independently of the loss of contrasts (e.g. Leda ~ leader) which caused it to emerge in non-rhotic dialects. I discuss these results in terms of sociophonology, which I use to convey the idea that the phonological process of hiatus-filling r-sandhi can spread through dialect contact, with a mixed phonological system emerging as a result. Although the data suggest a correlation between the loss of rhoticity and the development of r-sandhi, the nature of the overlap means that a phonological model must allow for speakers to have both features, even if rhoticity is eventually lost completely. Hay & Sudbury (2005) argue that the gradual development of linking and intrusive-r leading to their convergence to a single synchronic phenomenon ‘is not a process that can be well described by any categorical, phonological grammar’. I show that the current situation in East Lancashire speech can be described by existing phonological models with underlying representations and associated surface forms. These existing models do not rule out a parallel distribution for rhoticity and intrusive-r, in which individual speakers can have both features. This thesis provides some new dialectological data for an under-researched area of North West England, a discussion of phonological means of accounting for patterns in these data, and a discussion of the influence of socio-cultural spatiality on linguistic behaviour.