A forensic phonetic investigation of regional variation and accommodation in West Yorkshire

Abstract

This thesis presents an examination of regional variation and speech accommodation in two socially salient features of West Yorkshire English. The first aim of this research is to consider the extent to which local level variation exists across the West Yorkshire boroughs of Bradford, Kirklees and Wakefield. The second aim is to evaluate the effects of speech accommodation, the process whereby speakers adapt their speech production according to whom to they are talking (Giles, 1973; Giles & Powesland, 1975; Trudgill, 1981), in forensically-relevant contexts. The findings from these examinations inform how generalisable population data is for West Yorkshire across the three boroughs and also demonstrate to what extent accommodation could impact forensic speaker comparison (FSC) casework. The specific features examined in this thesis are the West Yorkshire FACE vowel and word-medial, intervocalic /t/. The motivations for examining these variables are twofold. Firstly, previous investigations of West Yorkshire English have suggested that these variables may be realised in different ways across the region and secondly, both variables appear to be socially salient in the speech community under investigation. As speech accommodation has been found to occur more often and to a stronger degree with respect to features that are socially salient (Cao, 2018; Smith & Holmes-Elliott, 2015; Trudgill, 1986), it was expected that the participants in this investigation would accommodate in respect of these speech parameters. However, the main focus of this investigation is to examine the magnitude and direction of any accommodation behaviour, and to evaluate the potential consequences this may have for FSC outcomes. This study is one of the first to make use of the newly published West Yorkshire Regional English Database (WYRED; Gold, Ross, & Earnshaw, 2018). The study analyses the speech of 30 males from West Yorkshire recorded completing three semi-spontaneous speaking tasks that utilise different interlocutors. Participants are equally split across the boroughs of Bradford, Kirklees and Wakefield and form a homogenous population in terms of age, gender and language background, enabling a systematic evaluation of regional variation. For the analysis of FACE, measurements are taken of the first three formants at 25%, 50% and 75% across the total vowel duration. Using these measurements, a series of statistical analyses are conducted in order to establish levels of variability across boroughs and across tasks. Additionally, realisations of intervocalic /t/ are analysed auditorily and assessments of variability between boroughs are carried out as well as an examination of changes in T-glottaling rates across tasks. For both speech parameters, accommodation is evaluated using an acoustic-phonetic approach whereby the participants’ realisations are considered in relation to those of their respective interlocutors. The findings of the investigations presented in this thesis reveal that FACE productions vary at the local borough level, specifically in terms of midpoint F2 values, whereas /t/ productions are not regionally stratified across West Yorkshire. Based on these results, recommendations are outlined for delimiting the relevant population for FSC casework involving West Yorkshire speakers. With regards to speech accommodation in FACE and /t/, results show that accommodation behaviour is highly variable across participants, both in terms of the direction and amount of accommodation present. All participants were considered to accommodate in at least one speech parameter, and a small number of participants displayed very high levels of within-speaker variability across tasks, highlighting the level of potential impact that speech accommodation can have on socially salient speech parameters. The consequences of these findings are addressed from both a FSC casework perspective and also in terms of sociolinguistic research practices more generally.