Lectal Focusing in Interaction

Abstract

A long-standing challenge in quantitative sociolinguistic analysis is identifying fine speaker meanings in interaction while retaining the ability to draw wider group comparisons. To bridge these goals, we propose a methodology for quantitative discourse analysis. In data from the Punjabi community in London, we initially find comparable group rates of use of an ethnolinguistic variable by older and younger British Asian (second-generation) men. We develop a new metric to assess whether these groups are in fact indexing similar ethnic and class meanings. Our measure of lectal focusing in interaction (LFI) tracks how much an individual shifts toward one or another style during a single interaction, focusing on Standard British English, Vernacular London English, and Indian English. Older British Asian men exhibit a high degree of LFI, shifting dramatically at times to achieve subtly strategic, interactionally tuned ends. Younger British Asian men show lower rates of LFI, particularly in their use of ethnic variants. Despite the continued use of similar forms, the LFI analysis identifies changes in indexical potential and a shift from marker toward indicator-like usage. We account for this through major changes in the social practices and political climate over recent decades in the community. The LFI measure thus brings interactional analysis to bear on the causes and rates of language change.

Keywords

discourse variation style ethnicity class indexicality interaction identity Indian English British English

Contemporary metropolitan regions such as London are major centers of migration and ethnic contact, with complex mixes of class and ethnic markers emerging in local speech. Take, for instance, Extract 1, from a phone conversation.

Extract 1: Anwar with Ronni

((Anwar, a middle-aged, middle-class British Asian Muslim businessman in Southall, West London, is chatting with his old school friend Ronni, a British Asian Sikh. They have just completed a business discussion. We hear only Anwar’s side of the phone call, with Ronni responding for one to two seconds between each of Anwar’s turns.))¹

1 Anw: tennu pata hai yaar

{you know how it is man}

2 Anw: hor kiddan? wha[ʔ]s goi[n] [ɖ]own man? every[f]i[ŋ] coo[w]?

{what else is up}

3 Anw: [h]ow’s [ø]ings a[ʔ] [ø]e yar[ɖ]?

4 Anw: [d̪]e o[l][ø] l[e][ɖ]y a[w]righ[ʔ]?

Anwar employs a complex mix of linguistic elements here: Punjabi language and London vernacular (glottaling, th-fronting, segmental deletion, and syntactic ellipsis) mixed in with Standard British English (–ing, h-retention) and even some Jamaican (yard) and other vernacular (old lady) vocabulary pronounced with Punjabi retroflexion.

Quantitative and qualitative methodologies in sociolinguistics offer distinct affordances in analyzing such speech. Quantitative approaches can support broader group depictions, but, due to the scale of the data, are somewhat removed from subtleties of social indexicality and speaker meaning. Qualitative approaches can explore the finer details of particular interactions, but tend to lose the ability to generalize outward from specific instances. While acknowledging the strengths of each practice for particular questions, researchers have long noted the need to explore the space between the two (e.g., Bell 2001), as important feedback between macro- and micro-level variation is otherwise rendered invisible.

Building on insights from both traditions, this study develops a new methodology to examine the meanings attributable to a speaker’s variable use of ethnic and class markers in speech. As a case study, we examine a British Asian community in West London that has experienced considerable social change over recent decades. We start with a broad quantitative observation, namely that older and younger British Asian men in West London share similar overall use of an ethnolinguistic variable, postalveolar articulation of /t/. We then devise a metric to track how interactionally tuned an individual’s variable use of this and other features is. This metric is quantitative but operates at the micro-level of interaction. We discover that despite broad similarities, the groups differ in how strategic their use is. Older men synchronize their use of variants closely to the unfolding interactional work at hand, whereas younger men’s usage, particularly of ethnolinguistic variants, is much less discernibly strategic in discourse. This shift in use rather than frequency corresponds to changes in the sociopolitical experience of different age groups within the second generation, and to the stabilization of a British Asian speech style.

The proposed method permits a finer understanding of continued or changing speaker meanings, deriving from speakers’ changing social realities in the community. We hope to contribute to recent efforts to bridge variationist and interactional methods in the study of indexical meaning in language variation. Ultimately, such micro-quantitative discourse analysis may also help to clarify the role of interaction in language change.

The Challenge: Inferring Meaning from Frequencies

The starting point for this study is a single question: particularly in situations of contact that involve highly mixed speech styles, does frequent use of a given speech form always imply strategic or agentive use? Intuitively, the answer to this question is “no”; sociolinguists have long known that speakers may be entirely unaware of very frequent features in their speech (e.g., Trudgill 1983), or, conversely, may be highly strategic in their use of very rare tokens of a form (e.g., Snell 2010). The question that follows is more challenging: how can we establish how strategic or agentive a speaker’s use of a form is?

To answer this, sociolinguists have often relied on close qualitative analysis. As we wish to generalize beyond this, to comment on group dynamics and community change, we explore a micro-quantitative measure that permits a degree of comparison across speakers. Although we hope that this approach brings elements of interactional analysis into variationist analysis and vice versa, our focus is on the former, namely enriching the study of language variation and change with close discourse detail. To this end, and due to limitations of space, our discussion of previous literature in this section is confined to the study of agency within quantitative variation rather than interactional research, and we focus on measures of speaker meaning and agency rather than social meaning more broadly.

Quantitative Measures of Interactional Meaning

Many variationist methodologies were developed to meet goals somewhat distinct from those of the present study. They explored the social embedding of language variation and change by correlating frequencies of use with different demographic segments of a speech community. Sociolinguists rarely claimed that such correlations offered a complete picture of social meaning. Indeed, they often furnished qualitative or attitudinal evidence to flesh out an understanding of social meanings where needed (e.g., Labov 1963, 1966). Nevertheless, frequencies of use have at times been used as a basis for inferring social meaning: “The traditional emphasis in variation studies has been to correlate linguistic variables with macro-sociological categories (e.g., class, gender, ethnicity), and to take the correlation to be a sufficient characterization of the variable’s social significance” (Eckert 2008b:26). Specialized social factors can be used to strengthen such inferences. Within the study of ethnicity, for instance, Hoffman and Walker (2010:37) devised a measure of ethnic orientation, not simply ethnic category, to infer that certain speakers “use overall rates to express ethnic identity.” Guy and Cutler (2011) developed a sophisticated measure of an individual’s deviation from a mathematically derived “baseline” style, to interpret divergences as performance of identity.

A considerable body of work has explored measures of intraspeaker variation that get closer to the level of interaction and discourse. From the earliest urban sociolinguistic studies, the presence of style variation according to formality or attention paid to speech for a given variable has been read as social awareness on the part of the speaker (Labov 1972:208). Particularly in situations of contact, Labov proposed a now well-established tripartite distinction, such that indicators, below the level of consciousness, may come to be transformed into markers or stereotypes, involving higher levels of awareness and thus greater correlations with formality. However, formality “was not intended as a general description of how style-shifting is produced and organized in every-day speech, but rather as a way of organizing and using the intraspeaker variation that occurs in the interview” (Labov 2001a:87). Indeed, the analysis will show that both macro-sociological correlations and formality correlations in interviews show no major differences between the two groups of interest—older and younger British Asian men—whereas finer interactional measures do.

Further work in the field has targeted speaker agency more specifically by moving beyond formality and exploring such factors as topic and interlocutor (e.g., Coupland 1980, 2001; Bell 1984, 2001; Rickford & McNair-Knox 1994; Schilling-Estes 1998, 2004; Labov 2001a; Levon 2009).² Indeed, in a finding parallel in some ways to the present study, Levon (2011) used topic analysis to show that similar pitch patterns in two groups of Israeli lesbians did not correspond to parallel speaker meanings for the form.

To some extent, topic has been more straightforward to operationalize as a “speaker agency” factor than more interactional elements of speech events, even though interactional and discourse organization have long been recognized as central to speakers’ variation. A number of recent studies have developed exploratory quantitative measures of micro-interactional sensitivity to variable forms. Levon (2009) coded topic and conversational frame, and the interaction between the two, to expand the scope of quantitative coverage of interactional detail. Podesva (2007) used the term “style clusters” to track multiple variables and their co-occurrence in interaction, an important property of the methodology proposed here as well (and a desideratum for style analysis in general; cf. Bell 2001:168). Eckert (2008b) proposed specific, interactional meanings for /æ/-raising and nonraising by demonstrating their correspondence with interactional shifts to excited, peer-oriented speech in girls’ interactions. Damari (2010) showed that use of a larger (L1-influenced) vowel space by a bicultural individual aligned with specific oppositional stances in interaction. Kiesling (2009) has quantified rates of use for phonetic variables relative to stance and related speech activities within single interactions, and Levon and Holmes-Elliott (2013) have similarly shown the central role of such stance types in phonetic variation in British speech.

The metric described later builds on many of these insights. It is designed specifically to track clustered alternations (or the lack thereof) in the use of competing dialect forms in a contact situation.

Speaker Meaning and Agency

Ahearn (2001:112-113) defines agency as “the socioculturally mediated capacity to act.” In particular, she proposes that agency is not “ontologically prior” but arises from “the social, political, and cultural dynamics of a specific place and time.” As the sociocultural context in the present case (described in the next section) has transformed over recent decades, individual agency in linguistic practice may vary considerably. The present study aims to identify and account for such variation.

Agency invokes a set of further, related constructs or continua, including awareness, consciousness, intentionality, and nonautomaticity. Although the literature reviewed briefly below is associated with these, the method we devise does not strictly address them. It is mainly designed to identify visible patterns of variation, tying these closely into the flow of discourse so that we can infer what kinds of indexical meanings might be carried by variants in a speaker’s unfolding talk. Our focus is largely on establishing degrees of speaker agency and types of speaker meaning in the use of ethnolinguistic and class variables. Note that these are narrower than social meaning, as even in the absence of any discernible speaker meaning or intention, a form may well carry social meaning in the minds of listeners.

Debates over deterministic and agentive explanations for variable language use have been a constant in variationist sociolinguistics. Research on ethnolinguistic traits has often focused on agency in indexing ethnic identity (e.g., Le Page & Tabouret-Keller 1985; Schilling-Estes 2004; Fought 2006). Speakers have been assumed to activate “different parts of their linguistic repertoires selectively in order to highlight particular aspects of their social identities (and to downplay others) in particular settings” (Doran 2004:96), and to “adopt and use these features strategically” (Hoffman & Walker 2010:59). By contrast, many variationist studies of change over time have emphasized the deterministic nature of outcomes, with more unconscious or automatic speech accommodation and exposure over time being the driving mechanisms (Giles 1973; Trudgill 1986, 2004, 2008; Goldinger 1998; Pickering & Garrod 2004).

The relationship between mechanistic and socially motivated dimensions of accommodation is complex. This was remarked upon as early as Bloomfield (1933) and is widely recognized in the study of speech accommodation (see also Yaeger-Dror 1993 and Auer & Hinskens 2005). Indeed, many scholars acknowledge that sociolinguistic variation involves “a set of resources that speakers deploy both intentionally and automatically in their day-to-day practice” (Eckert 2008b:26; see also Babel 2009). Similarly, in relation to code-mixing in contact situations in particular, Vertovec (2009:74) quotes Ballard (1994) in recognizing that “sometimes the process is due to purposeful selection and emphasis, and sometimes it is non-conscious or inadvertent.”

Benor (2010:173) emphasizes the importance of not assuming agency a priori, but developing appropriate methods to examine degrees of awareness and manipulation of ethnolinguistic forms. Fortunately, more and less strategic use of forms should correspond to distinct distributions in conversation, so the contrast is testable given a delicate enough metric. Strategic deployment of a variant involves use based on specific social affordances for the speaker, and so should be systematically aligned with shifts in interactional purpose, such as footing (Goffman 1981) or narrative structure. By contrast, relatively nonstrategic use of a variant should lead to less predictably aligned distributions of variants in discourse, as the presence of variants may simply reflect stochastic exposure in the input during learning, perhaps with a broad register association but no goal of linking forms to specific moments.

The metric we develop here tracks variable forms in continuous speech, measuring the degree of focusing in the style shifts of a given individual relative to his or her interactional meaning or purpose at any moment. Unlike some previous work, style-shifting is not modeled here as divergence from a default style or a “true” vernacular, but rather as focusing toward contrasting styles at different points in time.

Examining the speech of second-generation British Asian men from two age groups that share a comparable overall frequency in their use of one ethnolinguistic variant, we investigate whether this aggregate similarity implies genuine similarity in use. We find systematic differences in the quality of indexical work done by ethnic and class variants for these individuals despite similar overall rates. Older men have more strategic, agentive use of Indian variants; over time, this leads to sustained use rather than avoidance of these pronunciations in the community, which forms the basis of continued but less finely controlled use among younger individuals.

The rapprochement between qualitative and quantitative analysis presented here in the form of quantitative micro-analysis aims to steer a path between the two pitfalls of macro-structural overdetermination on the one hand and a vacuous “multiple identities” fluidity on the other. This in turn permits a finer understanding of language change in the community. In terms of Labov’s three-way distinction in degrees of speaker awareness, the findings point to a shift from marker toward indicator status for many ethnolinguistic variants in the community. In terms of Auer’s (1999) language contact model, the contrast resembles a shift from code-mixing toward a fused lect.

The West London Punjabi Community

Background

This study focuses on the Punjabi community in Southall, West London. As a diasporic, lower-middle-class, Asian-majority suburb of London, the community involves a complex layering of ethnic and class speech features. The community is particularly useful for examining change in the valuation and use of such markers as it is one of the oldest South Asian communities in the United Kingdom, with sizeable first, second, and third generations.

In the present study we examine speech from selected older (ages thirty-five to fifty) and younger (ages eighteen to thirty-five) second-generation men. Although these individuals were all born and raised in West London, older and younger individuals grew up in very different sociopolitical climates. Over the course of sixty years, Southall has shifted from having a minority to a majority Asian population. Not unrelated to this, race relations have shifted from overt and violent hostility to cooperative coexistence. Sharma (2011) and Sharma and Sankaran (2011) outline details of this transition, providing quotes from the two age groups that reveal experiences of antagonism and hostility among older British Asians (see also CARF 1981; Meads 1983; Cashmore 1996; Oates 2002) in contrast to experiences of wider cultural acceptance among younger British Asians. Figure 1 summarizes these broad contrasts experienced by second-generation Southall residents at different points in history.

Figure 1.

Historical community context.

Semiethnographic fieldwork was conducted over a period of nine months by two fieldworkers (both female, Standard Indian English speakers). For the wider project, a total of seventy-four participants were recorded twice, along with multiple self-recordings in diverse settings collected by ten participants (two each from five demographic groups) in the absence of either researcher. Of these self-recordings, those conducted by older and younger second-generation men are relevant to the present study. In total, approximately 120 hours of data were collected. Extracts discussed in the present work are taken from interviews and self-recordings.

Variationist Measures of the Data

Figure 2 presents a simple quantitative comparison of the use of one ethnolinguistic variant in our Punjabi London data, namely articulation of /t/ beyond the alveolar range of British English. This postalveolar Asian range of articulation, incorporating retracted and retroflex variants, derives from a series of retroflex stops in Indic languages and Indian English. Previous work has found this feature to be present in the speech of younger second-generation British Asians, primarily in the form of retracted rather than retroflex variants (Alam & Stuart-Smith 2011; Kirkham 2011). These finer phonetic distinctions are extremely important in understanding finer social meanings. Nevertheless, in the present study we group them together as broadly Asian, similar to our single grouping of a range of vernacular London English diphthong variants. We do this in part because we are focusing on broad ethnic and class contrasts, and also for feasibility, as the study looks at multiple variants of thirteen variables using auditory analysis (discussed later in relation to Table 1).

Figure 2.

Use of [ʈ] in interviews by older and younger British Asian men and women.

Table 1.

Coding of Lectal Focusing in Interaction (LFI).

	Standard BrE	IndE	Vernacular BrE
Variables coded
postsonorant, word-final and intervocalic (t)	t, tʃ, ɾ	ʈ	ʔ, ø
postsonorant, word-final and intervocalic (d)	d, dʒ	ɖ	ø
coda and syllabic (l)	ɫ	l	w
goat diphthong	əʊ, oʊ, ə	o	aʊ
face diphthong	eɪ	e	aɪ
voiceless interdental fricative	θ	t̪^h	f, ø, t
voiced interdental fricative	ð	d̪	v, ø, d
word-final, root/suffix -ing	-ŋ	-ŋg	-n
noncluster, word-initial /t/	t	ʈ	—
noncluster, word-initial /d/	d	ɖ	—
Isolated phonetic forms	linking /r/	v/w alternation	h-dropping

Figure 2 (twenty-four individuals, 5,540 tokens) indicates that older and younger men behave rather similarly, both in terms of overall rates of use and the direction of a formality effect (confirmed as a mild but significant effect in a regression analysis in Sharma & Sankaran 2011).³ We might infer from this that men in this community share a similar use of this ethnolinguistic variable regardless of age; we might even conclude that the two male age groups express or index their ethnicity similarly, given the absence of the sharp decline found for women.

But is this comparison of frequencies sufficient to draw such conclusions? Do older and younger men vary in similar ways across contexts or in discourse? Is the intended meaning always similar? More precisely, is a given form always signaling ethnicity (group level), or could it be indexing finer meanings pertaining to scene or stance (individual and interactional level)? Might variants operate at different orders of indexicality (Silverstein 2003; Johnstone & Kiesling 2008) for different individuals, even if those individuals appear similar in their overall rates of use? We know from the outline history of the community that these age groups have experienced starkly different social conditions. Could this amount to distinct “sociocultural mediation of their capacity to act,” in Ahearn’s (2001:112) terms, and if so, can we find any evidence of such differences?

Our examination of men’s variation (or lack thereof) during the course of interactions and narratives will reveal that, despite all initial appearances, these men are in fact doing strikingly different things with ethnolinguistic and class variants. Importantly, these differences are not idiosyncratic to the individual. They appear to relate to ongoing sociopolitical, historical, and linguistic change in the community.

Methodology: Measuring Lectal Focusing in Interaction (LFI)

As noted, orders of indexical meaning can vary over time (Johnstone & Kiesling 2008). An individual might ascribe very fine interactional functions to a variant in interaction, or they might pay little attention to the presence of a form within their speech. The former targets a subtle meaning and the latter perhaps a very coarse or generalized group association (or none at all, if entirely below the level of consciousness). To identify the relative granularity of meaning, we need to track the amount and type of style shifting in a speaker’s discourse, noting whether variable features are finely tuned to interactional stance or not. This sort of measure can begin to point to why features are used by a given individual.

The data used include narratives selected for being among those with the highest emotional engagement evident in both older and younger men’s interviews, to minimize differences in style arising simply out of differences in degree of affect expressed. Themes that tended to elicit such engagement included racial tension and amusing personal narratives, both of which are therefore included. In addition to narratives, we include a sample of cross-interlocutor extracts for the older man in particular, to further illustrate his range of variability. We do not include such a set of interlocutor extracts for younger men for two reasons in addition to space constraints: younger men did not exhibit comparable variation across interlocutors, and intrasituation variation (compared in narratives from the two age groups) is more important for accessing a finer level of strategic use or nonuse.

The methodology we use is a simple metric to track what we term “lectal focusing in interaction” (LFI).⁴ The LFI measure offers a simple proportional measure of fluctuation in style over the course of a segment of interaction.

First, a given extract is segmented into units. For the quantitative tracking of variation, moderate sized units are important. If too small, units will face excessive skewing due to low numbers, and if too large, averaging over the unit may obscure internal variation. So the primary criterion for segmenting is major clausal boundaries, which tends to generate reasonably sized units for the present analysis, with a denominator of approximately ten tokens. Other analyses of variation in interaction have dealt with units at the level of the token (Kendall 2013), the utterance (Podesva 2007), and the topic (Schilling-Estes 2004; Levon 2009). As the goal of the measure in our study is to track fluctuations in style during interactions, we also attend to turn-constructional units (TCUs) and footing shifts as secondary criteria. Footing shifts are noted through marked shifts in pitch, volume, voice quality, topic, addressee, voicing, and alignment, among other factors (Goffman 1981:128). We describe these as secondary rather than primary criteria because relying exclusively on footing shifts can lead to some very long units, which can obscure variation, and relying exclusively on TCUs can lead to many small units, skewing average values. Thus, we first segment the text into major clausal units, and then add unit boundaries if these units include the end of a turn or a marked footing shift. As the metric does not aggregate measurements, effects occurring across larger chunks of discourse, such as footing or topic, are easily captured, as they would be visible as steady patterns maintained over groups of units.

Next, each unit is coded for a set of variables. For our present purposes, we group variants into recognized, enregistered lects in the community (cf. Wells 1982; Pingali 2009): Standard British English (SBrE), Vernacular British English (VBrE), and Indian English (IndE). A similar metric could track individual variables; indeed, this would be a more data-driven approach, acknowledging that each variable may have a distinct indexical field (Eckert 2008a). This might be preferable in situations where broad indexical values of variables are unclear. Our particular interest is style-shifting that potentially invokes macro-social ethnic (British versus Indian) and class (standard versus vernacular) indexicalities in the community, and so we believe the analysis benefits from lectal groupings of variants.⁵ However, we remain as conservative as possible in our choice and classification of variants into lects, excluding more ambiguous cases.

For the most part the Principle of Accountability is observed in our coding decisions.⁶ Primarily variables that show clear contrasts among the three lects are coded. As the coding is auditory, we follow the common practice of carving continuous phonetic space into discrete variants.⁷

Table 1 lists eight variables with a ternary contrast across the three varieties, as well as two salient ethnic contrasts that involve a binary rather than ternary distinction, contrasting along the British/Indian dimension but not along the Standard/Vernacular BrE dimension. The resulting absence of two VBrE variants skews coding very marginally toward SBrE. Some of the variants for these ten variables are very rare, such as VBrE null realization of selected forms in connected speech. The lower half of Table 1 lists three features, one for each lect, that are only coded when isolated but stylistically marked occurrences arise. This decision accommodates Bell’s (2001:167) observation that “[r]eferee design will often deal in the qualitative, the one-off, the single salient token which represents an identity.” Similarly, the use of a word, phrase, or grammatical construction clearly associated with IndE or VBrE was also coded. The remaining uncoded text consists of variables that are either not audibly contrastive across the lects or not reliably codable with auditory analysis.⁸

In the final step, a simple proportion is calculated for each of the three lects per unit, dividing the number of variants coded for each lect by the total number of variants coded in a given unit.

To help illustrate this procedure, the appendix includes the complete LFI calculation for Extract 2 (the first extract analyzed in the next section). We describe here how the first unit in this extract is coded. The eight-word unit opens with a Punjabi greeting (hor kiddan), a rare instance in which a lexical form is counted because it clearly contributes to a more Asian interactional style. In the remaining text of unit 1, we can identify five of the phonetic variables noted in Table 1 and one more clear vernacular lexical choice (man) that contributes to a London vernacular style.⁹ Each of the five other variants (initial /d/, intervocalic /θ/, coda /l/, and two instances of -ing) is coded by both authors as closest to one of the three variants listed in Table 1. Finally, a percentage value is calculated for each lect, the numerator adding up the number of forms found for a given lect and the denominator being the total number of variants coded. In the analysis that follows, a graph of the LFI calculation for each extract is also provided.

Several basic challenges remain in this exploratory approach. First, because the measure balances fine detail with quantitative generalization, it does not at present take into account the variable influence of internal factors. For instance, Sharma and Sankaran (2011) showed that position in the word influenced the likelihood of retroflex /t/ differently in the older and younger generations, so a preponderance of particular contexts in a given unit may skew the average for that unit. The LFI measure could eventually factor this information in and weight variants accordingly. For now, we avoided coding one context known to have a strong skewing effect, namely adjacent homorganic obstruents. In the case of the –ing variable, the literature on American English (e.g., Labov 2001b) indicates a disfavoring effect of noun status on –in realization; however, in keeping with recent findings for London English (e.g., Schleef, Meyerhoff & Clark 2011:235), our data did not indicate a strong effect of this type, and we include a full range of –ing forms. For other variables, such as /l/-quality, findings vary across studies. Our earlier examination of /t/-glottaling did show an effect of position in word (Sharma & Sankaran 2011), but these are known to correspond to degrees of vernacular meaning, and so are included. From a practical point of view, as we look at thirteen variables participating in clustered shifts, the likelihood of one internal factor in one variable skewing results is lessened. We do not build in detailed weightings based on internal factors at present, but we present all coding details in the appendix.

Second, variants contribute to lectal meaning with differing degrees of strength and specificity. This seems evident in the present data: for instance, we will see that Cockney diphthongs are relatively rare, especially among younger men, and occur mainly in heightened instances of stylization or sharp style-shifting. By contrast, deletion of final /d/ in and is commonplace in casual speech. Although both are treated as broadly contributing to VBrE style, they are clearly very distinct. Again, weightings could be built into an LFI measure, but for now we avoid adding this layer of manipulation.

Finally, we opted to conform to the Principle of Accountability as consistently and reliably as possible, and so omitted innumerable subtler phonetic shifts that tended to cluster with the coded lectal variants, for example, aspiration, dentalization, consonant lengthening, and many more segmental and prosodic variants.

In the analysis that follows, we first examine interactional and narrative extracts from an older second-generation man, and then compare these to narrative extracts from the speech of younger second-generation men. Although only individual case studies can be reported here to illustrate the full details of generational differences in LFI, independent evidence supports our interpretation of these as broad generational changes. These include systematic age differences in the influence of social factors (Sharma & Sankaran 2011) and in individual repertoires (Sharma 2011), as well as ethnographic observation in the community over an extended period.

Analysis: LFI over Time in the British Asian Community

Older Men

We start with Anwar—the speaker in Extract 1—as a case study. To begin with, we examine his interactions with four different interlocutors. In all four extracts, the speech activity is very similar—inquiring after the interlocutor’s well-being—yet Anwar modifies his direction and degree of LFI considerably for different interlocutors. The last of these four extracts, as well as a fifth and final narrative extract, show that Anwar’s LFI is not simply echoing his interlocutor but actively shaping meanings within his discourse.

For readability, we provide a simple orthographic transcription of each extract and a graph of the corresponding LFI measurement. In all examples, numbers are assigned only to speech units of the individual being analyzed. Details of coding for each example (bracketed IPA notation for all coded phonetic variables and underlining for all lexical and grammatical forms) are supplied in the appendix.

The first example—Extract 2—is a longer extract from the conversation presented earlier in Extract 1 and was discussed as an illustrative example of LFI coding in the previous section.

Extract 2: Anwar to Ronni, Asking after Family

((one- to two-second replies by Ronni between each turn))

1 Anw: hor kiddan what’s goin’ down man everything cool?

{what else is up}

2 Anw: how’s things at the yard?

3 Anw: the old lady alright?

4 Anw: you’re not giving her any trouble, are you?

5 Anw: yeah, you better behave yourself man kick your arse in otherwise

((discussion of problems, some in Punjabi))

22 Anw: just leff it man just leav- le- just leff it if she says right about this just give her what her dues are

23 don’t bloody drink anything don’t spend any money on the booze and drink

24 this is gonna take you down man

25 and we’re not gonna get any younger I’m telling you this right

26 we’re forty-two years old now y’know

The graph in Figure 3 presents the LFI measures for Extract 2. Including three separate lectal lines quickly becomes visually unwieldy in longer extracts. We therefore present all graphs in the form of two lines: a primary and a secondary measure. The primary measure, the solid line, is the ethnolinguistic contrast: the solid line in Figure 3 (and all remaining figures) tells us what proportion of coded forms in each unit were BrE (both Standard and Vernacular) as opposed to IndE. The higher this line, the more a BrE style is employed, and the lower it is, the more an IndE style is used. The secondary measure, the broken line, indicates class styles: it tells us what subset of the BrE forms were Vernacular BrE forms—when the solid and broken lines overlap, the British forms are all vernacular, and when the dotted line is much lower than the solid one, the British forms are predominantly standard. It is important therefore to first read the solid line, and then interpret the broken line as providing a further breakdown of that BrE usage.

Figure 3.

LFI in Anwar to Ronni, “Asking after Family.”

In Figure 3, we see regular fluctuation between BrE and IndE variants (the solid line). The close tracking of the solid line by the broken line indicates that the overwhelming majority of BrE forms are vernacular rather than standard style. The interaction is therefore overwhelmingly composed of a VBrE base with substantial contributions from IndE and further elements of Punjabi and London Jamaican style (yard, leff it).¹⁰ There is almost no role for SBrE. Two segments of the interaction are included to show the sustained use of this style. Anwar is fully aware of his selective use of this multiethnic style, describing it as “Southallian,” a code he says he uses with school friends of his generation, suggesting a degree of enregisterment.

Now compare Figure 3 to Figure 4 and the corresponding text in Extract 3. Here, Anwar is conducting the same interactional work as he was with Ronni in Extract 2—shifting from business matters to family. In Extract 3, however, Anwar is addressing an upper-middle-class British Asian Muslim barrister who uses an exclusively standard/posh phonetic range.¹¹ We see an entirely different lectal balance here, with an overwhelming use of SBrE in Figure 4. As with Extract 2, later segments of the interaction showed steady maintenance of the style, in this case majority SBrE forms, with slight fluctuations in the low rates of VBrE and IndE.

Figure 4.

LFI in Anwar to Bilal, “Asking after Family.”

Extract 3: Anwar to Bilal, Asking after Family

1 Anw: and he -s wants me to, you know, ehm be recipient of his eh moneys and

funds that he gets from royalties and eh and sponsorship

2 Anw: uhm and eh we may need eh your services in preparing some

documentation for any

Bil: yeah I- I can knock up a power of attorney (property) document there’s

no problem about that at all

3 Anw: yeah okay bu- I’ll keep you informed in what’s happening

4 Anw: ↑hows everything else? how’s the family?

Bil: yep alhamdulillah very very good actually. ah I’ve got a latest addition

I’ve got got a daughter ↑how many kids d’you have now?

5 Anw: eh I’ve got eh two daughters and eh one eh son

A third, very brief example is included in Extract 4 simply to demonstrate Anwar’s range of styles for this one speech activity. Here again Anwar is asking after the general well-being of his interlocutor; in this case, however, it is a working-class Sri Lankan maid and we see a total shift to IndE by Anwar.¹² In addition to the core features coded for LFI, we highlight other forms, to clarify the complete lectal shift, including vowels, absence of aspiration, and consonant lengthening. This is a wholesale dialect shift with no remnant of SBrE or VBrE (a chart would simply show the IndE line at 100 percent and both BrE lines at 0 percent throughout, in stark contrast to Figures 3 and 4). At times in his interview with Devyani (first author), Anwar also executed near-complete shifts to IndE, some shown later in Figure 6.

Extract 4: Anwar to Rosa, Asking after Health

Anw: h[æ][l][o] yes eh h[æ][l][o] Rosa ho[v] are you? you ø [o][k][e]↑? yes you [k]ee[p:]ing [v]e[l]?

The contrast among Extracts 2, 3, and 4 is very clear: Southallian to Ronni, SBrE to Bilal, and IndE to Rosa. This clear interlocutor effect supports, at the very least, marker-like sensitivity in his use of style clusters to different Asian interlocutors. We might conclude that Anwar’s use of class- and ethnically marked variants is nothing more strategic than a broad speech accommodation effect.

The two remaining examples from Anwar’s speech—one dialogic and one monologic—militate strongly against this reading. Both illustrate his use of more fine-tuned LFI within speech directed to a single interlocutor, showing highly controlled use of variants for micro-interactional work.

First, to illustrate variation at a finer level than interlocutor but still dialogic, consider Extract 5 and Figure 5, in which Anwar greets his interlocutor, and then requests help with a problem. Anwar’s addressee here is a mechanic, Ishfaq, a working-class Asian Eastender with an accent as traditional Cockney as Bilal’s is posh. (Indeed, Anwar remarked that he recorded this phone call on speakerphone to show how impossible it would be to guess from his voice that Ishfaq is in fact a British Asian Muslim.)

Figure 5.

LFI in Anwar to Ishfaq, “Reason for Calling.”

Extract 5: Anwar to Ishfaq, Reason for Calling

Ish: how you doin’ bruv?

1 Anw: ye:ah I’m fine thanks how you doin’? y’ okay?

Ish: y- yeh not too bad bruv

2 Anw: y’know ehm eh th- this these eh insurance people they’re really m-

mucking me around right [now-

Ish: [s’ what they sayin’?

3 Anw: well you know what I mean they’re just pussy-footing about, they are, you

know?

Ish: hhahahahahahaha hehehehehehehe

4 Anw: y’know eh ss so ((smiley voice)) so- listen how we gonna get this car

sorted out man?

When speaking to the mechanic, Anwar starts with a mix of Standard and Vernacular BrE, and the first time he announces his problem (unit 2) Ishfaq’s response is noncommittal. Instead of responding to Ishfaq by providing the requested details, Anwar just recodes his general sense of grievance in more of the London vernacular (units 3-4), introducing h-dropping, Cockney diphthongs, and more vernacular discourse markers. This time Ishfaq bursts into a hearty laugh. Anwar’s reformulation is no more informative than his first attempt, so it must be the way he says it that captures Ishfaq and engages him.

Anwar is not simply returning his accent to that of his interlocutor in a broad sense. Instead, as the talk unfolds, he is using classed speech forms to shift the footing and adjust his interactional demeanor.

The final example from Anwar’s recordings demonstrates that this fine manipulation of class features also extends to ethnic features. Extract 6 is from an interview narrative, so is more monologic than previous examples.

Extract 6: Anwar to Devyani (Interviewer), Museum Visit Narrative

and um so- this is- is- that is a brilliant splitting hairs na-

don’t don’t forget this country is a very notorious country let’s not forget

that these people are premeditated they are premeditated conspirers

they have divided our country and they have ruled in our country

they have done the disgraceful acts. they have- they have massacred

they have made each other, they have orchestrated.. each others- they have orchestrated massacres

I’m not talking about now but I am talking about the ideology is still within their mindset, you know

India. the greatest biggest massacre that happened. Muslims killing Sikhs, Sikhs killing Muslims, you know

you know it was a turmoil

an- who orchestrated it

you ask anybody now who orchestrated it

the British orchestrated it and the British people are doing the same

you see they are dividing and ruling

even here, look within us, they are dividing the business community with the residents

you see it’s a divide and rule policy. it is in their- their core

and you’re not gonna get away from that and we have to stand up beyond that

this is why I’m always tolerant you know

some- we were invited to the er Royal Albert Museum

and they said look you know you community leaders you are you know we want Zou to- invite you to the British heritage

and I went to the- I went there and I said

aw that’s beautiful that’s lovely. aw look at that

the Elgin marbles are there oh look at that mosque. the member of the mosque it-

the m- MEMBER y’know the member where the- where the where the minister sits you know

you- they-v- you have raped the mosque. you have taken it out. you put it here.

this shouldn’t be here it should be in a mosque in Turkey

now elgin marbles. they should be in gr- in in er in Greece. you know

they shouldn’t be here so I came out

and they said oh yes sir sir how did you enjoy your trip?

aw fantastic it’s wonderful

and what do you think I said you really want to know what I think

it w- a warehouse of stolen goods you know and that created uproar

I said that was a warehouse of stolen goods and I’m ashamed to be British.

after I went into the v and a Victoria and Albert museum

this’s what I feel

In Figure 6 we see dramatic fluctuations in Anwar’s overall use of ethnic variants (the solid line) and in his use of classed BrE forms (the broken line). We see moments of 100 percent use of BrE variants but also moments of 100 percent use of IndE variants. These fluctuations help to structure and voice the unfolding narrative. Of particular interest is Anwar’s reliance on IndE lectal focusing to convey a range of affective stances—particularly personal and political outrage (units 3-6, 13-15) and cultural insult (units 23-24)—and discourse moves such as the response segments of rhetorical question-response structures (units 11-12). Many of these can be interpreted as uni-directional (Bakhtin 1984; Rampton 1995), such that he identifies with the values conveyed by the lectal voice. He adopts SBrE lectal focusing for narrative framing moves (units 2, 9, 30) and for a closing “moral high ground” evaluation of the experience (units 30-35), indicated in the widening separation of the solid and broken line as the narrative closes. Remarkably, he employs vernacular, Cockney-inflected BrE stylization to cast the voice of a gullible provincial guest awed by the museum’s riches (unit 21). Arguably, this voice is closest to his own demographic—a British man raised in a working- and lower-middle-class neighborhood of London in the 1970s—and yet this is the voice that he marks as most “othered” in this particular interaction.

Figure 6.

LFI in Anwar to Devyani (interviewer), “Museum Visit” narrative.

Many further phonetic details feed into this structured variation. For instance, in his pronunciation of member twice in units 22-23, Anwar alters the quality of his /r/, consonant length in /mb/, and the vowel in the final syllable to shift from SBrE to IndE, as he shifts from neutral observation to shock.

Of course, Anwar is audibly the same person across all the extracts presented here, drawing on a single pool of linguistic features: Punjabi, IndE, varieties of Vernacular London English, Standard BrE, and even occasionally Creole-influenced London Jamaican forms. Even so, it is clear that (a) he turns some of these linguistic elements up and others down as he moves from one addressee to the next, (b) that he turns elements up and down even within conversations, relative to footing shifts and narrative structure, and (c) he is very reflexive about these interactional sensitivities, referring to “Southallian” in the interaction with Ronni, describing speech like the barrister’s as “polished,” and saying that the mechanic is a Cockney, a “thoroughbred Eastender . . . of Pakistani origin.”

This LFI sketch of Anwar reveals a landscape of variation that incorporates two standard varieties and several vernaculars. His movement across this range suggests continual reinscription of specific social and ethnopolitical commitments. Variants are clearly not arbitrarily distributed in Anwar’s speech: Extracts 5 and 6 in particular showed fine interactional and narrative focusing of multiple lects. These preclude a description of his variation as simply a stochastic mix that derives from exposure to different variants during his lifetime. Space prohibits the inclusion of more data, but many other episodes in Anwar’s and several other older men’s recordings show this skilled tuning of variants to discourse work.

It is worth noting a secondary point, namely that a more specific relationship between interlocutors and discourse meanings also emerges. As noted, Anwar shows substantial attunement to specific interlocutors. Each of these interlocutors appears to set boundaries of variation, within which different lectal sets of variants have distinct meaning potentials and are played up or down as the interaction unfolds. These lects do not have default indexical values that transcend contexts: we do not see stance meanings for lects that persist across interactions, such as Standard BrE = negative politeness, IndE = mockery, or Vernacular BrE = assertiveness (cf. Ochs 1992). Indeed, stance values are so heavily constrained by interlocutor that we even see entirely inverse indexical functions with different addressees. For instance, in the interaction with Devyani (first author), Anwar used Cockney as an explicitly mocking and “othered” voice (Extract 6, unit 21), but in his interaction with Ishfaq, he used the same features in an appeal for solidary understanding (Extract 5, unit 3). At least at the level of clustered lects, indexical meaning potentials appear to be highly circumscribed by Anwar’s orientation to interlocutors. We explore reasons for this in the closing discussion.

We have focused here on close analysis of one individual but our dataset indicates similarly high fluctuation in LFI in the speech of many middle-aged men in this community. Sharan, the other older second-generation man who provided self-recordings, showed similar multidialectalism (see Sharma 2011) and shifted sharply between lects in interactions with clients in his shop.

Next, we ask whether younger men, who appeared similar to older men in terms of overall frequency of use of an ethnic form, show the same types and degree of LFI.

Younger Men

Anwar’s data indicated two levels of LFI: intersituational and intrasituational. A careful examination of the younger men’s interactions turned up no instances of ethnolinguistic lectal focusing to match Anwar’s degree of LFI, either intersituationally or intrasituationally. At the intersituational level, Sharma (2011) showed that, unlike the robust multidialectal alternation of older men, not one of the sixteen younger men in the project showed this sharp style of alternation. Despite retaining many of the same ethnolinguistic variants, younger men showed a very restricted range of fluctuation between situations, at least in terms of ethnolinguistic style (not necessarily in terms of standard-vernacular British variation). Since intrasituational LFI can reveal finer strategic use, we turn here to whether intrasituational variation also shows a restricted range among younger men.

The two extracts that follow are, like Anwar’s Extract 6, high involvement narratives from interviews. We examine whether younger men mirror Anwar’s wide range of lectal manipulation within a single situation, seen earlier in Figure 6. One extract relates an experience of racial tension, like Anwar’s narrative, and the other involves humor and traditional Punjabi cultural knowledge, themes that also elicited dramatic LFI increases in IndE style in Anwar’s interview.

The first narrative—presented in Extract 7 and Figure 7—was produced by Ravinder, a nineteen-year-old, lower-middle-class law student. Comparing the solid line in Figure 7 to the solid line earlier in Figure 6, we see that although IndE variants are present in Ravinder’s speech, they do far less work in Ravinder’s narrative than in Anwar’s. The presence of IndE does not rise above the level of 30 percent (units 1 and 3), in contrast to Anwar’s regular instances of 100 percent IndE style (Extract 4; Extract 6, units 11 and 14). Furthermore, an examination of specific variants (in the appendix) shows that the fluctuation corresponds more to fixed usage—certain variants (e.g., th-stopping) are consistently Asian, others not—than to stylistic sensitivity across all variants, as in Anwar’s usage.

Figure 7.

LFI in Ravinder to Devyani (interviewer), “Phone” narrative.

Extract 7: Ravinder to Devyani (Interviewer), “Phone” Narrative

((Devyani has asked whether Ravinder considers his neighborhoods safe.))

no I used to live in a safe neighborhood but over here this th- som- some problems happening there

cos like once I was I went to cinema to see my friends and um er there’s three group of boys there, there’s some black and some white

and they come up- they came up to me and they go ‘are you Asian?’ i said ‘yea’

and they go ‘so gimme your phone gimme your phone then’ because I’m Asian

something like that. so I didn’t really give- I walked off

and then um my friends were there as well so lucky nothing happened

but that’s the (xxx) if I said I’m black they wouldn’t have taken my phone

but then white, they would’ve they’d’ve let me go

The second narrative, in Extract 8 and Figure 8, was produced by Anand, who, like Ravinder, is a lower-middle-class, twenty-two-year-old law student.

Figure 8.

LFI in Anand to Lavanya, “Bhangra Team” narrative.

Extract 8: Anand to Lavanya, “Bhangra Team” Narrative

we’re sort of a s- we started off as amateurs but made out as if we were professionals to go into the- (xxx) we were like ºyea yeaº we got a gig

I remember our first rehearsal, there were only four of us and there was one dholi {drummer}

and we’re like aw the other eight guys are doing a gig in Birmingham so they couldn’t make it

but this is for xxx ((anonymized)) we’ll do the rehearsal anyway, us four will do it

we did a two minute routine (mini) minor one with a few good individual moves but not like a routine where we’re all mingling that much

and they’re like “yea you’re wicked you’re amazing, let’s get you in the act”

so the rest of you going to be there next month?

yep okay ((smiley voice)). no routine, nothing and we had a month to go till um we performed in the xxxxx ((anonymized))

like aw crap, we better do something now. right guys guys let’s start rehearsing, hhheh, let’s come up with a routine. we’ve only got a month

and we did it. we practised every day for an hour or two

and I think the week before the gig we practiced three hours heh everyday until we got it right

and like ((laugh)), this is it. we’re ready now. (xxx) heh pulled it off, did a fifteen minute routine. eheh

As with Ravinder’s usage, Anand shows steady presence but relatively low variability in his use of IndE forms, only twice exceeding the 30 percent level slightly (units 5 and 7). In both narratives, it is difficult to identify clear discourse functions for tiny variations. Even if we assert that the peak in units 6 and 7 is discourse-linked, the fluctuation is very small. These narratives also show a corresponding lack of sustained shifts to SBrE, another contrast to Anwar’s usage.

It would be wholly inaccurate to say that, given their lower LFI along the ethnic dimension, Ravinder and Anand simply build less dynamic tension into their narratives than Anwar does. It is their choice of devices that differs: both younger men show greater fluctuation in their use of BrE class-marked variants—65 percent in both cases—than ethnic variants, exploiting a typical British standard-vernacular range to structure their narratives. Like Anwar, their use of SBrE style corresponds at times to orienting and evaluation phases of the narrative (e.g., Extract 7, unit 1; Extract 8, unit 8), which are possibly less “personalized” or more “objectivized” segments (Gumperz 1982). They also make extensive use of typical monolingual devices common among their non-Asian peers, such as shifts in tempo, timing, and volume (Gardner-Chloros, Charles & Cheshire 2000).

In sum, even though Figures 7 and 8 are among the most variable that we can find in our data from younger men, they come nowhere near the degrees of LFI fluctuation found for Anwar and his age group. Where older men’s use of Asian variants is alert to ethnopolitical valuation or “acts of identity,” illustrated in particular detail in Anwar’s alternations in Extract 6, the younger men’s use appears to exploit such orientations less, possibly operating at a lower level of awareness, as part of a “fused lect” (Auer 1999), thus potentially indicative of a more advanced stage of lectal formation.¹³ Anand even voluntarily raised the question of intentionality when discussing his own speech, shown in Extract 9.

Extract 9: Anand (Younger Man)

when I’m with my um my Punjabi peers . . . every now and then a word or two in Punjabi will come in, but we intend that to happen. It’s intentional. and then there’s other times when it happens unintentionally with um my English friends . . . I’ll speak an English word but it’ll come out with an Indian accent.

Conclusions

The LFI metric has helped unravel some of the tangle of features observable in British Asian speech. Beneath a highly diverse surface, with a similar overall mix of features shared across the men in the community, underlying generational differences emerged through the LFI analysis. Though simple and exploratory, the methodology contributes further to the recent interest in the multivalency of variables (Bucholtz 2009; Wong & Hall-Lew 2012), as well as to the question of how indexicality and interaction might relate to language change.

In initial comparisons of older and younger second-generation British Asian men, group averages and formality effects in the use of an ethnic variant were nearly identical. However, the LFI analysis has shown that ethnic and class variants behave differently in the two groups. The analysis of Anwar (and observation of parallel individuals) suggests that older men show regular deployment of such variants for finely tuned interactional work. The meanings produced by these uses sometimes evidenced participation in distinct social scenes (Extracts 3-5), but sometimes simply exploited variants to highlight discourse shifts (e.g., Extract 6, units 10-12, units 16-18), suggesting that not all uses of ethnolinguistic traits correspond to a broad “ethnic identity” meaning (all Anwar’s interlocutors are Asian, yet his range of variation is dramatic). Despite comparable overall rates of use, younger men seemed less delicate and agile in the interactionally tuned distribution of their lectal variants, particularly ethnolinguistic forms.

We can explain these generational differences by returning to the community history outlined in Figure 2. Older men born in Southall grew up as a minority in a hostile cultural climate, and spent their lives negotiating two often starkly separated ethnic groupings. They participated actively in their Asian cultures, often going into their fathers’ businesses, marrying women from South Asia, and maintaining substantial personal and business contacts with India, all of which required signaling, and ensured exposure to, “authentic” Asianness. They were also part of potentially hostile local school environments and local British political scenes, which required signaling “authentic” Britishness. These experiences certainly underpin their remarkable native-like command of multiple lects, which plays an important part in making these available for discourse work. A keen sense of bicultural reality permeates their interviews, finding expression in, among other things, the kinds of dramatic shift in interactional stance and alignment associated with the wide fluctuations in Anwar’s LFI. We can now speculate that stylistic dispositions of the type displayed by Anwar arise from the specific social climate that formed the backdrop for his generation. His particularly complex ethnopolitical agenda and productive sensitivity to macrosocial ideologies of British ethnic and class positioning embody what Bourdieu (1985:728) describes as a “practical mastery of the social structure.”

Younger men grew up as a local majority in a far more accepting and less polarized climate. They have much weaker ties to India, following few of the social practices of the older British-born generation. They have also experienced much less politicized local British lives, and their limited LFI—a lower range of use of ethnic variants than class variants for discourse work—may well signal the emergence of a more unified, rather than bicultural, British Asian identity (Harris 2006).

In terms of language change and consciousness, this change corresponds to stages of long-term lectal focusing (Trudgill 2004; Auer 1999), with a shift from marker-like use among older men toward indicator-like use among younger men. Although this appears to run counter to the commonly cited transformation of indicators into markers through contact (Labov 1972), it is simply a late stage of the same trajectory: indicator (precontact IndE variant) > marker (socially variable use in early stages of contact with BrE) > indicator (less variable use in a more “mature” British Asian lect).

What theoretical insights can we draw from these findings? Eckert (2009:21) observed that “stylistic activity runs the entire gamut from quite automatic accommodation to completely contrived performances, and all of these offer their own kind of evidence of the social meaning of variation.” The LFI metric has helped us propose very different reasons for similar overall presence of traits at different stages in the community: earlier more strategic uses, and later less interactionally sensitive use. In one sense, this suggests different types of ethnic identity at different stages. The older men mark ethnic positionings in moment-to-moment interaction, maintaining a sense of the distinctiveness of different social realms; the younger men inhabit a less politicized, yet recognizably ethnic British identity. The indexical value of variants changes despite their continued use. These changes in value make visible very subtle social change in the community.

Remaining challenges and complications of the LFI methodology were noted earlier. Nevertheless, the methodology may help to bring qualitative analysis of interaction to bear on the dynamics of language change. First, in terms of causes of change, we see that, if they are unqualified by closer discourse analysis, the direct extrapolation of meanings from quantitative distributions can risk ascribing the wrong meaning to variants (e.g., masculine or Asian), which can lead to an inaccurate account of the motivation of a change. Second, in terms of rate of change, LFI analysis can help explain the longer retention of exogenous traits by one ethnic group as opposed to another by identifying the social work such forms do (or don’t do). In the present case, the older men’s politicized choice to maintain a substantial, though carefully calibrated, ethnolinguistic styling has led to continued exposure to and acquisition of certain forms by younger second-generation men, albeit with a loss of some of the sociopolitical commitments and indexicalities.

Footnotes

Appendix

Acknowledgements

We are very much indebted to the other members of the project team, Lavanya Sankaran, Pam Knight, and Roxy Harris. We also thank Karel Arnaut, Sarah Benor, Deborah Cameron, Lars Hinrichs, Scott Kiesling, Malcah Yaeger-Dror, Jaspal Singh, Amelia Tseng, and audiences at numerous presentations for comments that improved earlier versions.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Economic and Social Research Council, UK (“Dialect Development and Style in a Diasporic Community,” RES-062-23-0604).

Notes

Author Biographies

Devyani Sharma is reader in sociolinguistics at Queen Mary University of London. She works on sociolinguistics, language contact, bilingualism, and syntactic variation, with a particular interest in new varieties of English. Her books include Research Methods in Linguistics (Cambridge 2013, co-edited with Rob Podesva).

Ben Rampton is professor of applied and sociolinguistics and director of the Centre for Language Discourse and Communication at King’s College London (). He does interactional sociolinguistics, and his interests cover urban multilingualism, ethnicity, class, youth, and education. His books include Crossing: Language and Ethnicity among Adolescents (Longman 1995/St Jerome 2005).

References

Ahearn, Laura

2001. Language and agency. Annual Review of Anthropology 30. 109-137.

Alam

Farhana

. 2011. Identity and ethnicity in /t/ in Glasgow-Pakistani high-school girls. In Proceedings of the 17th International Congress of Phonetics Sciences (ICPhS), 216-219. Hong Kong, China.

Auer

Peter

. 1999. From codeswitching via language mixing to fused lects: Toward a dynamic typology of bilingual speech. Journal of Bilingualism 3(4). 309-332.

Auer

Peter

Frans Hinskens. 2005. The role of interpersonal accommodation in a theory of language change. In Auer

Peter

Hinskens

Frans

Kerswill

Paul

(eds.), Dialect change: Convergence and divergence in European languages, 335-357. Cambridge: Cambridge University Press.

Babel

Molly

. 2009. Phonetic and social selectivity in speech accommodation. Berkeley: University of California, Berkeley dissertation.

Bakhtin

Mikhail

. 1984. Problems in Dostoevsky’s poetics ( Emerson

Caryl

, ed. & trans.). Minneapolis: University of Minnesota Press.

Ballard

Roger

. 1994. Desh Pardesh: The South Asian presence in Britain. London: Hurst.

Bell

Allan

. 1984. Language style as audience design. Language in Society 13(2). 145-204.

Bell

Allan

. 2001. Back in style: Reworking audience design. In Eckert

Penelope

Rickford

John R.

(eds.), Style and sociolinguistic variation, 139-169. New York: Cambridge University Press.

10.

Benor

Sarah Bunin

. 2010. Ethnolinguistic repertoire: Shifting the analytic focus in language and ethnicity. Journal of Sociolinguistics 14(2). 159-183.

11.

Bloomfield

Leonard

. 1933. Language. London: Unwin.

12.

Bourdieu

Pierre

. 1985. The social space and the Genesis of groups. Theory and Society 14(6). 723-744.

13.

Bucholtz

Mary

. 2009. From stance to style. In Jaffe

Alexandra

(ed.), Stance: Sociolinguistic perspectives, 146-170. Cambridge: Cambridge University Press.

14.

CARF (Campaign Against Racism and Fascism). 1981. Southall: The birth of a black community. London: Institute of Race Relations.

15.

Cashmore

Ernest

. 1996. A dictionary of race and ethnic relations. London: Routledge.

16.

Coupland

Nikolas

. 1980. Style shifting in a Cardiff work setting. Language in Society 9. 1-12.

17.

Coupland

Nikolas

. 2001. Language, situation and the relational self: Theorizing dialect-style in sociolinguistics. In Eckert

Penelope

Rickford

John R.

(eds.), Style and sociolinguistic variation, 185-210. New York: Cambridge University Press.

18.

Damari

Rebecca Rubin

. 2010. A phonetic analysis of stancetaking by a binational couple. Southampton, UK: Sociolinguistics Symposium 18 paper.

19.

Doran

Meredith

. 2004. Negotiating between “Bourge” and “Racaille”: Verlan as youth identity practice in suburban Paris. In Pavlenko

Aneta

Blackledge

Adrian

(eds.), Negotiation of identities in multilingual contexts, 93-124. Clevedon: Multilingual Matters.

20.

Eckert

Penelope

. 2008a. Variation and the indexical field. Journal of Sociolinguistics 12(4). 453-476.

21.

Eckert

Penelope

. 2008b. Where do ethnolects stop? International Journal of Bilingualism 12(1). 25-42.

22.

Eckert

Penelope.

2009. Three waves of variation study: The emergence of meaning in the study of variation. http://www.stanford.edu/~eckert/thirdwave.html (20 February 2012).

23.

Fought

Carmen

. 2006. Language and ethnicity. Cambridge: Cambridge University Press.

24.

Gardner-Chloros

Penelope

Charles

Reeva

Cheshire

Jenny

. 2000. Parallel patterns? A comparison of monolingual speech and bilingual code-switching. Journal of Pragmatics 32(9). 1305-1341.

25.

Giles

Howard

. 1973. Accent mobility: A model and some data. Anthropological Linguistics 15(2). 87-105.

26.

Goffman

Erving

. 1981. Forms of talk. Philadelphia: University of Pennsylvania Press.

27.

Goldinger

Stephen D.

1998. Echoes of echoes? An episodic theory of lexical access. Psychological Review 105. 251-279.

28.

Gumperz

John

. 1982. Discourse strategies. Cambridge: Cambridge University Press.

29.

Guy

Gregory

2011. Speech style and authenticity: Quantitative evidence for the performance of identity. Language Variation and Change 23(1). 139-162.

30.

Harris

Roxy

. 2006. New ethnicities and language use. London: Palgrave Macmillan.

31.

Hinrichs

Lars

. 2011. The sociolinguistics of diaspora: Language in the Jamaican Canadian community. Austin: SALSA XIX, University of Texas keynote address.

32.

Hoffman

Michol

James

Walker

. 2010. Ethnolects and the city: Ethnic orientation and linguistic variation in Toronto English. Language Variation and Change 22(1). 37-67.

33.

Johnstone

Barbara

Kiesling

Scott F.

2008. Indexicality and experience: Exploring the meanings of /aw/-monophthongization. Journal of Sociolinguistics 12(1). 5-33.

34.

Karp

Ivan

. 1986. Agency and social theory: A review of Anthony Giddens. American Ethnologist 13(1). 131-137.

35.

Kendall

Tyler

. 2013. Speech rate, pause, and sociolinguistic variation: Studies in corpus sociophonetics. London: Palgrave Macmillan.

36.

Kiesling

Scott Fabius

. 2009. Style as stance. In Jaffe

Alexandra

(ed.), Stance: Sociolinguistic perspectives, 171-195. Oxford: Oxford University Press.

37.

Kirkham

Sam

. 2011. The acoustics of coronal stops in British Asian English. In Proceedings of the 17th International Congress of Phonetics Sciences (ICPhS), 1102-5. Hong Kong, China.

38.

Labov

William

. 1963. The social motivation of a sound change. Word 19. 273-309.

39.

Labov

William

. 1966. The social stratification of English in New York City. Washington, DC: Center for Applied Linguistics.

40.

Labov

William

. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.

41.

Labov

William

. 1982. Building on empirical foundations. In Lehmann

Winfred P.

Malkiel

Yakov

(eds.), Perspectives on historical linguistics, 17-92. Amsterdam: John Benjamins.

42.

Labov

William

. 2001a. The anatomy of style-shifting. In Eckert

Penelope

Rickford

John R.

(eds.), Style and sociolinguistic variation, 85-108. Cambridge: Cambridge University Press.

43.

Labov

William

. 2001b. Principles of language change, vol. 2, Social factors. Malden, MA: Blackwell.

44.

Le Page

Robert

Tabouret-Keller

Andrée

. 1985. Acts of identity: Creole-based approaches to language and ethnicity. Cambridge: Cambridge University Press.

45.

Levon

Erez

. 2009. Dimensions of style: Context, politics and motivation in gay Israeli speech. Journal of Sociolinguistics 13(1). 29-58.

46.

Levon

Erez

. 2011. Teasing apart to bring together: Gender and sexuality in variationist research. American Speech 86(1). 69-84.

47.

Levon

Erez

Holmes-Elliott

Sophie

. 2013. Style over substance: /s/-variation and stance-taking in southeast England. Pittsburgh: NWAV 42, Carnegie Mellon/University of Pitts-burgh paper.

48.

Meads

R. J.

1983. Southall 830-1982. London: Merlin.

49.

Oates

Jonathan

. 2002. Southall and Hanwell. London: History Press.

50.

Ochs

Elinor

. 1992. Indexing gender. In Duranti

Alessandro

Goodwin

Charles

(eds.), Rethinking context: Language as an interactive phenomenon, 335-358. Cambridge: Cam-bridge University Press.

51.

Pickering

Martin J.

Garrod

Simon.

2004. Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences 27(2). 169-226.

52.

Pingali

Sailaja

. 2009. Indian English. Edinburgh: Edinburgh University Press.

53.

Podesva

Rob

. 2007. Three sources of stylistic meaning. Austin: SALSA XV, University of Texas paper.

54.

Rampton

Ben

. 1995. Crossing: Language and ethnicity among adolescents. London: Longman.

55.

Rickford

John & Faye McNair-Knox.

1994. Addressee- and topic-influenced style shift: A quantitative sociolinguistic study. In Biber

Douglas

Finegan

Edward

(eds.), Sociolin-guistic perspectives on register, 235-276. New York: Oxford University Press.

56.

Schilling-Estes

Natalie

. 1998. Investigating “self-conscious” speech: The performance register in Ocracoke English. Language in Society 27(1). 53-83.

57.

Schilling-Estes

Natalie

. 2004. Constructing ethnicity in interaction. Journal of Sociolinguistics 8(2). 163-195.

58.

Schleef

Erik

Meyerhoff

Miriam

Clark

Lynn

. 2011. Teenagers’ acquisition of variation: A comparison of locally-born and migrant teens’ realisation of English (ing) in Edinburgh and London. English World-Wide 32(2). 206-236.

59.

Sharma

Devyani

. 2011. Style repertoire and social change in British Asian English. Journal of Sociolinguistics 15(4). 464-492.

60.

Sharma

Devyani

Sankaran

Lavanya.

2011. Cognitive and social factors in dialect shift: Gradual change in London Asian speech. Language Variation and Change 23(3). 399-428.

61.

Silverstein

Michael

. 2003. Indexical order and the dialectics of sociolinguistic life. Language and Communication 23(3/4). 193-229.

62.

Snell

Julia

. 2010. From sociolinguistic variation to socially strategic stylisation. Journal of Sociolinguistics 14(5). 630-656.

63.

Trudgill

Peter

. 1983. On dialect: Social and geographical perspectives. Oxford: Blackwell.

64.

Trudgill

Peter

. 1986. Dialects in contact. Oxford: Blackwell.

65.

Trudgill

Peter

. 2004. New-dialect formation: The inevitability of colonial Englishes. Edinburgh: Edinburgh University Press.

66.

Trudgill

Peter

. 2008. Colonial dialect contact in the history of European languages: On the irrelevance of identity to new-dialect formation. Language in Society 37(2). 241-254.

67.

Vertovec

Steven

. 2009. Transnationalism. London: Routledge.

68.

Wells

J. C.

1982. Accents of English. 3 vols. Cambridge: Cambridge University Press.

69.

Wong

Amy

. 2012. Regional variability and ethnic identity: Chinese Americans in San Francisco and New York City. Portland, OR: Linguistics Society of America paper.

70.

Yaeger-Dror

Malcah

. 1993. Linguistic analysis of dialect “correction” and its interaction with cognitive salience. Language Variation and Change 5(2). 189-224.