Overview

The English spoken in Boston has long been characterised by non-rhoticity, but a gradual return to rhoticity has been observed over the last few decades (Wells 1982; Wolfram and Shiling-Estes 2006; Staun 2010). However, a renewal of the empirical data and discussions seems much needed as only few recent studies have offered systematic analyses of rhoticity in the region (Nagy and Roberts 2004; Irwin and Nagy 2007; Nagy and Irwin 2010).

In this chapter, we wish to discuss the results of a survey conducted in Boston within the PAC programme (Durand and Przewozny 2012; Brulard et al. 2015). We briefly describe our post-Labovian approach to data collection as well as our annotation and coding method using Praat and Dolmen for the extraction of the coding information. Our data have been analysed auditorily and an /r/ coding system has been implemented for ten speakers who showed variable rhoticity (the remaining three showing stable non-rhoticity) amounting to a total of 3,050 tokens. This coding system takes into account factors such as the position of /r/ within the syllable (onset or coda) and its phonological environment, as well as the stress level and quality of the pre-/r/ vowel for coda-/r/.

From a quantitative standpoint, our results show that rhoticity in Boston is highly variable, with individual rates of coda-/r/ realisation ranging from 23 per cent to 93 per cent, and it seems register-sensitive since wordlists yield higher rates of realisation than spontaneous speech tasks. Rhoticity is also more frequent in final position than in pre-consonantal position which echoes the classical findings of Labov (1966) in New York City. Stress proves to be a relevant factor since the frequency of realisation of coda-/r/ increases with the level of stress of the syllable. Our analyses of pre-/r/ vowels are in line with the results of Irwin and Nagy (2007) as we find words of the NURSE lexical set to favour /r/ realisation dramatically (99 per cent) while other vowels yield results around 65 per cent, with the exception of schwa whose unstressed status disfavours /r/ realisation.

Although we have not investigated r-intrusion systematically in variably rhotic Bostonian speakers, we have found a few cases of intrusive ‘r’ in the connected speech tasks of our protocol.