This article reports a study of acoustic phonetic variation between ethnic groups in the realisation of the British English liquids /l/ and /ɹ/. Data are presented from 'Anglo' and 'Asian' native speakers of Sheffield English. Sheffield Anglo English is typically described as having 'dark' /l/, but there is some disagreement in the literature. British Asian speakers, on the other hand, are often described as producing much 'clearer' realisations of /l/, but the specific differences between varieties may vary by geographical location. Regression analysis of liquid steady states and Smoothing Spline ANOVAs of vocalic-liquid formant trajectories show consistent F2ࢤF1 differences in /l/ between Anglo and Asian speakers in non-final contexts, which is suggestive of a strong distinction between varieties in terms of clearness/darkness. There is also evidence of a polarity effect in liquids, with differing relationships between liquid phonemes in each variety: Asian speakers produce /l/ with higher F2ࢤF1 values than /ɹ/, and Anglo speakers produce /ɹ/ with higher F2ࢤF1 values than /l/. The results are discussed in terms of phonetic variation in liquids and socioindexical factors in speech production.

A B S T R A C TThis article examines how the social meanings of phonetic variation in a British adolescent community are influenced by a complex relationship between ethnicity, social class, and social practice. I focus on the realisation of the HAPPY vowel in Sheffield English, which is reported to be a lax variant [ɛ] amongst working-class speakers but is undergoing change towards a tense variant [i] amongst middle-class speakers. I analyse the acoustic realisation of this vowel across four female communities of practice in a multiethnic secondary school and find that the variable's community-wide associations of social class are projected onto the ethnographic category of school orientation, which I suggest is a more local interpretation of class relations. Ethnographic evidence and discourse analysis reveal that local meanings of the HAPPY vowel vary further within distinctive community of practice styles, which is the result of how ethnicity and social class intersect in structuring local social practices. (Intersectionality, indexicality, social meaning, identity, ethnicity, social class)*

This article investigates how variation across different levels of linguistic structure indexes ideological alignments in political talk. We analyse two political speeches by Ed Miliband, the former leader of the UK Labour Party, with a focus on the use of /t/-glottalling and the types of verb processes that co-occur with the pronouns we and you. We find substantial differences in the production of /t/ between the two speeches in words such as Britain and government, which have been argued to take on particular salience in British political discourse. We contextualise these findings in terms of metalinguistic discourse surrounding Miliband's language use, as well as how he positions himself in relation to different audiences via verb process types. We show that phonetic variation, subject types, and verb processes work synergistically in allowing Miliband to establish a political persona that is sensitive to ideological differences between different audiences. (Social meaning, indexicality, political discourse, verb processes, phonetic variation, /t/-glottalling)*

This article investigates the acoustic and articulatory correlates of vowel contrasts in bilingual speakers. We analyse data from bilingual speakers of Twi (Akan) and Ghanaian English, with the aim of examining how the production of the advanced tongue root vowel contrast in Twi relates to the production of the tense/lax vowel contrast in Ghanaian English. These data are compared to tense/lax vowel data from monolingual British English speakers. The acoustic results show that Twi and Ghanaian English mainly rely on F1 for distinguishing [ATR] and [TENSE] vowels, whereas British English uses F1, F2, F3 and duration for the [TENSE] contrast. The ultrasound tongue imaging data show tongue root distinctions across all languages, while there are consistent tongue height distinctions in British English, no height distinctions in Ghanaian English, and small height distinctions for some vowels in Twi. Twi has the weakest correlation between F1 and tongue root advancement, which suggests that the [ATR] contrast may involve additional strategies for pharyngeal cavity expansion that are not present in [TENSE] vowels. In doing so, we show that bilinguals produce similar contrasts in similar ways across their two languages, but that language-specific differences also persist, which may reflect different articulatory goals in each language.

literature. Jones (1966, 92) speculates that Liverpool /l/ may be clear in all positions, 61 stating that 'its existence there is probably due to Irish influence', with many varieties of 62 Irish English having very clear /l/s. Knowles (1973, 256) claims that /l/ in Liverpool is 63 frequently 'velarised' and produced in similar ways across positions. One of the few sources 64 of instrumental data on Liverpool /l/ comes from Turton (2014), who reports acoustic 65 and ultrasound data on a single male speaker. She finds that he produces the initial∼final 66 contrast in /l/, but that he also produces word-final /l/ with distinct velarisation, as opposed 67 to the more pharyngealised articulations documented for other British English varieties. This 68 also suggests a potentially 'intermediate' realization for Liverpool /l/, which may lie towards 69 the middle of a continuum between clear and dark. 70In this study, we address the relationship between time-varying lateral and vowel formant 71 dynamics. Accordingly, we briefly overview previous research on vowels in each variety. 72Manchester English shows features typical of many northern Englishes, such as the lack of a 73 foot-strut or trap-bath split and monophthongal productions of canonical diphthongs 74 (Baranowski and Turton, 2015). Liverpool English typically merges the nurse and square 75 vowels (Knowles, 1973; Watson, 2007) and has complex patterns of raising in price and 76 mouth before nasal-obstruent clusters (Cardoso, 2015). 77A concrete difference between dialects that we predict will have an effect on our results is 78 the final vowel in words such as belly (Wells 1982 calls this the happy vowel). Manchester is 79 reported to produce very low and back variants of happy (Baranowski and Turton, 2015), 80 which we do not expect to see in Liverpool. Finally, we discuss pre-lateral vowels, which are 81 5 Dialect variation in formant dynamics particularly significant for our study. Fronting of /u/ is typically inhibited before coda /l/ in 82 some varieties of English (Kleber et al., 2011), although the articulatory interpretation of this 83 is not straightforward (Strycharczuk and Scobbie, 2017). However, Baranowski (2017) finds 84 a clear social class effect on pre-lateral /u/ fronting in Manchester, with a strong negative 85 correlation between social class and fronting in this context. While we are not aware of any 86 studies of pre-lateral /u/ in Liverpool, our own impressions suggest that fronting of /u/ 87 before coda /l/ is widespread in this dialect. 88 B. Time-varying spectral analysis 89 The significance of the time-varying properties of sonorant sounds has been comprehen-90 sively documented in the literature (Elvin et al., 2016; Fox and Jacewicz, 2009; Strycharczuk 91 and Scobbie, 2017; Watson and Harrington, 1999; Williams and Escudero, 2014). This is 92 particularly pertinent to a study of laterals, which are inherently non-static due to the tim-93 ing relations outlined in Section I A, as well as the existence of strong interactions be...