Tag Questions and Indexicality

Tag questions are well suited for exploring the connection between micro-, meso- and macro-social meanings, since the relationship between their functional properties (their ability to mark stance) and their propensity for indexing styles and social categories has long been recognized. Lakoff (Reference Lakoff1975:15), for example, argues that tags – “midway between an outright statement and a yes-no question” – suggest a lack of confidence on the speaker’s part. She points out that this perceived lack of confidence at the micro-social level could come to index a social type at the macro-social level, arguing that women are expected to use tentative forms, which simultaneously subordinates them and marks them as feminine. However, the categorical labeling of questions as less assertive than statements is, of course, problematic when one takes the context of an utterance into account. Compare, for instance, a statement like I’m not much of a dancer uttered by an especially shy adolescent with the question Do you want to take this outside? uttered by a burly nightclub bouncer.

Many sociolinguists have drawn attention to the multiple pragmatic functions tag questions can serve, faulting Lakoff for focusing on just one.Footnote ³ Holmes Reference Holmes1982, Reference Holmes1984, Reference Holmes1995 was the first to seriously draw attention to the manifold purposes that tags can serve. She distinguishes between epistemic modal tags (those that request information about which the speaker is genuinely unsure, e.g. Nobody’s using this chair, are they?), and affective tags (those that indicate concern for the addressee either by being facilitative, e.g. It’s a lovely day, isn’t it?, or softening, e.g. That was a bit silly, wasn’t it?). Based on a quantitative analysis of her corpus, Holmes reports that epistemic modal tags and softening affective tags are used more by men, and that facilitative affective tags are used more by women. Crucially, she illustrates that tags may not only express a lack of confidence, but can be used to facilitate discourse or interpersonal relationships.

Others have gone on to propose a wider range of stances potentially articulated by tag questions. Drawing on Holmes’s work, Cameron, McAlinden & O’Leary Reference Cameron, McAlinden, O’Leary, Coates and Cameron1989 point out that tag questions can be used to request information, increase an utterance’s politeness, or facilitate talk. Importantly, and in contrast to Lakoff Reference Lakoff1975, not all of these functions work to subordinate women. Algeo Reference Algeo1988 makes a similar argument, asserting that in British English tags can sometimes be used in a peremptory or aggressive manner. He suggests that tags are employed for such purposes by women and men alike, and further notes that neither type of tag is likely to be interpreted as sounding tentative. Finally, Winefield, Chandler & Bassett Reference Winefield, Chandler and Bassett1989 also argue against the deferential interpretation of tag questions. Although their quantitative analysis reveals that a female patient’s use of tag questions increases over the course of psychotherapy, qualitative analysis reveals that her specific use of tags represents increased levels of independence.

While taking stock of the pluralistic pragmatic functionality of tags, it is important to acknowledge that determining such functions is by no means a straightforward endeavor. Holmes (Reference Holmes1995:113) admits that “interpreting the function of tags is a subjective business”; the need to add a third category of affective tag, “challenging,” to her 1995 analysis alludes to the difficulty in assigning affective meanings to the tag questions in her analysis. Furthermore, it is not necessarily the case that using tags to articulate one stance precludes the simultaneous conveyance of another. As Coates (Reference Coates1987:130) observes, speakers regularly “exploit the polypragmatic nature of [linguistic items] to say many things at once.” Accessing speakers’ motivations in particular moments of interaction, to the extent that it can be done, presents the sociolinguist with a difficult challenge. Consequently, we argue that we also stand to benefit from attending to what the structure of the tag enables a speaker to communicate.

Cheshire Reference Cheshire1981 attends to precisely this issue in her examination of the interplay between the grammatical composition of a tag and its function in the discourse. Examining tags relative to the wider linguistic practice of the speakers who used them, she distinguishes between tags whose content could be confirmed/disputed (because the interlocutor knew the information) and those whose content could not (because the interlocutor did not have the required knowledge). The former, which she calls “conventional tags,” occurred with a wide range of verb forms, from nonstandard ain’t and int to standard negated forms (e.g., ‘isn’t’). In contrast, nonconventional tags occurred only with int. Cheshire further observes that these int tags achieved an aggressive meaning and were used predominantly by adolescents engaged in the vernacular culture. Here then, the meaning of the tag question as aggressive and/or challenging is determined not only by Cheshire’s “sense” of the stance articulated by the tag, but also by virtue of its grammatical design. This particular tag was composed using a marked nonstandard form associated with the aggressive vernacular style in which it predominantly occurs. Cheshire’s (Reference Cheshire1981) work highlights our earlier point that single linguistic features like tag questions are stylistically embedded and may indeed express meaning through other linguistic features, such as the morphosyntactic forms they contain. For Cheshire, the structure of the tag (which was indexically linked to a social style) disambiguated its social meaning.

In our analysis of tag questions, we aim to present a stylistic analysis of the trajectory of social meaning that connects micro-social stances to macro-social social types, via meso-social personae. We accomplish this task first by utilizing quantitative methods to identify the basic and generalizable function of tag questions, which we discern via an analysis of the discourse contexts in which the variable occurs. We then trace how this feature is used to articulate a range of stances that are distinctly styled in the interactions of different communities of practice. In adhering to style, our analysis does not just assess grammatical structure, but also considers the phonological shape of tag questions, their turn context and design (where a tag occurs in a speaker’s turn and the response it achieves), and the nature of the talk in which they occur. The study differs from previous quantitative work on tag questions in that it locates statistical trends in the social practices and identities of its speakers. Other quantitative approaches (Andersen Reference Andersen1998, Reference Andersen2001; Nordenstam Reference Nordenstam1992; Tottie & Hoffman Reference Tottie and Hoffmann2006), though very robust for the purposes of identifying statistical trends, draw on corpora that lack ethnographic detail. We conclude by exploring how meaning is constructed recursively across linguistic and social contexts, endeavoring to illustrate the multimodal (i.e., constructed in social and linguistic practice) and indexically layered (i.e., adapted and redesigned across ideological space) nature of social meaning.

The Study

The data for this study are drawn from Moore’s (Reference Moore2003, Reference Moore2004) ethnography of female students at Midlan High, a school in northwest England. Between June 2000 and February 2002, Moore engaged in participant observation with a group of 40 girls in Year 8 (aged 12 to 13) when the research began and in Year 10 (aged 14 to 15) on completion of the fieldwork. Throughout this period, Moore went into the school to spend the lunch hour with the girls and engaged in whatever activity was occurring (this could range from simply eating lunch in the canteen, to attending a dance practice, to hanging out with the smokers behind the sports hall). The fieldwork was documented in field notes amounting to 196,400 words. Interactions between the girls were recorded from Year 9 onward. Recordings never took place as a formal interview and served as an extension of Moore’s regular interaction with the girls. The girls would discuss recent events and used the sessions as a way to record their activities. In addition to these recordings, questionnaires were also circulated to the study’s participants. These questionnaires provided an informant view on social practices, and were used to corroborate Moore’s ethnographic observations.

The recordings comprise approximately 50 hours of data and a 262,000-word corpus. In all, the corpus includes 778 tag questions. In this number, we count only those tags that follow a declarative. While McGregor Reference McGregor, Hasan and Fries1995, Reference McGregor1997 offers a typology of tag questions including those that can be preceded by interrogatives, imperatives, and exclamatives, forms other than declaratives were either exceptionally rare or nonexistent in the corpus. With respect to the form of the tag itself, we examined only those tag questions consisting minimally of both a verb and a subject. Lexical tags (e.g., huh, right) were thus excluded. We also excluded invariant verbal tags (e.g., innit). Although this grammaticalized feature predominates in London youth culture (Andersen Reference Andersen2001), our recordings from northwest England contained a single token of invariant innit (Cos it depends what mood she’s in, innit?). It appears that although the feature is widespread in the boroughs of London, it has not taken hold as strongly in the North, at least not among the adolescent girls studied here.Footnote ⁴

Given that linguistic features are used with reference to an overall linguistic and social style, the girls in this study were categorized according to community of practice (CofP) membership (Eckert & McConnell-Ginet Reference Eckert and McConnell-Ginet1992). Each girl’s CofP membership was determined on the basis of whom she spent time with, what activities she engaged in, her orientation to peers and her surroundings, personal style and appearance, and her own and others’ assessments of her group memberships. The four CofPs identified over the course of fieldwork can be superficially positioned on an anti-school to pro-school continuum. This continuum, though something of a generalization, is typical of school ethnographies that categorize students on the basis of their levels of engagement in activities deemed illegitimate by the school, such as smoking, truancy, and unauthorized occupation of space (cf. Willis Reference Willis1977). Figure 1 depicts this orientation and summarizes the number of speakers in and tags produced by each group. The Populars exhibit an anti-school attitude, display a sporty and feminine style of dress, and engage in moderately rebellious activities like drinking and smoking. They often associate with boys of the same age, frequently as their “girlfriends.” Midway during the fieldwork, the TowniesFootnote ⁵ emerged and broke off from the group of Populars, when the former started to engage in risky behavior such as drug-taking and sexual activity. Their friendship network includes older boys, with whom they spend time in the local area. Townies have a more streetwise dress style, which includes wearing gold jewelry and heavy makeup. The Geeks, by contrast, maintain most of their relationships through the school in activities like orchestra and sport. They wear little makeup and jewelry and take a more practical approach to their appearance. Although the Geeks, like the Populars, associate with male peers, Geeks form exclusively platonic relationships with boys. Finally, the Eden Village girls adopt a trendy “teen” style, wearing pastels and glittery accessories, preferring to spend free time dancing, shopping, and at sleepovers. Their friendship network is rather close-knit and excludes boys.Footnote ⁶

FIGURE 1:

Communities of practice at Midlan High

Each of the 778 tag questions was coded for the tagger’s CofP, as well as a number of discursive, grammatical, and phonological factors. At the level of discourse, we carefully examined the discourse in determining (i) whether interlocutors expressed agreement with the tag; (ii) whether interlocutors overlapped with the tagger’s turn, and if so where; and (iii) whether tags occurred at the end of a turn or turn-medially. We examined agreement (i) primarily because much has been made of the solidarity function of tags (Coates Reference Coates1987; Holmes Reference Holmes1984, Reference Holmes1995). We were particularly interested in quantitatively examining whether the use of tags does in fact result in explicit expressions of accord. Each tag was categorized as being met with assent (i.e., the interlocutor explicitly agrees with the proposition expressed by the tagger), dissent (i.e., interlocutor explicitly disagrees), no response (i.e., interlocutor makes no reference to proposition expressed by tagger), and other (i.e., idiosyncratic responses, such as a request for clarification or an ambiguous response). We coded for overlap (ii) in order to investigate the potential role of tag questions in organizing discourse, marking whether interlocutors overlapped with the tagger before, during, or after the tag, if at all. We also examined the turn context (iii), coding each tag as either turn-final or turn-medial.Footnote ⁷ If tags encourage interlocutors to verbally engage in the interaction at hand, then we would expect tags to be placed at the ends of turns, where they facilitate turn transition.

Given Cheshire’s (Reference Cheshire1981) discovery that the internal composition of tags influenced their ability to communicate social meaning, we additionally coded for a number of grammatical and phonological factors. At the grammatical level, we examined the grammatical subject of the tag, as previous research has noted the high frequency of tags with it (Tagliamonte Reference Tagliamonte1998, Tottie & Hoffmann Reference Tottie and Hoffmann2006). The great majority of subjects in this study were pronouns; we distinguished between each, including the distinction between generic and specific you. Studying grammatical subject also enabled us to consider the type of talk in which tag questions occur, as we discuss later. Also at the grammatical level, we considered whether tags included nonstandard morphosyntax, enabling us to identify any correspondences between tag design and the use of nonstandard forms. By “nonstandard forms,” we refer to fully lexical forms, namely nonstandard were (the most common nonstandard form of agreement in the corpus) and ain’t. Here we made a three-way distinction between tags with nonstandard morphosyntax (e.g. His mum were downstairs, weren’t she?), those with standard morphosyntax (where nonstandard forms would have been possible; e.g. They’re happy, aren’t they?), and those in which there was no environment for nonstandard forms (e.g. She can be alright, can’t she?). The latter were excluded from this part of the analysis.

At the phonological level, we examined the realization of word-final /t/ within tag questions. We included this variable both because of the feature’s social salience (informants remarked on their parents’ “corrections” of intervocalic [ʔ] in the discussions of language use convened by Moore post-data collection) and because of the great deal of attention the glottalized and released variants of the region have received in the sociophonetics literature, particularly with respect to “youth norms” (Docherty, Foulkes, Milroy, Milroy & Walshaw Reference Docherty, Foulkes, Milroy, Milroy and Walshaw1997; Foulkes, Docherty & Watt Reference Foulkes, Docherty and Watt1999, Reference Foulkes, Docherty and Watt2005; Llamas Reference Llamas2007; Milroy, Milroy, Hartley & Walshaw Reference Milroy, Milroy, Hartley and Walshaw1995; and various articles in Foulkes & Docherty Reference Foulkes and Docherty1999). We were able to consider this feature because the overwhelming majority of the tags in our corpus exhibited negative polarity, the most common tag form in English (Kimps Reference Kimps2007), and hence carried a contracted form of not. Each word-final /t/ appearing in a tag question was coded as released, glottal(ized), deleted, or palatalized,Footnote ⁸ and the effect of linguistic factors (preceding and following phonological context) was also considered.

In addition to the aforementioned features, we investigated the verb used in the tag, syllable deletion, and the dropping of word-initial /h/. As none of these analyses revealed strong patterns according to CofP membership, we do not discuss them further here. One potentially salient phonological feature of tag questions not considered here is intonational shape. Although many scholars have taken note of the pragmatic importance of intonation to the interpretation of tag questions (Ladd Reference Ladd1981, Millar & Brown Reference Millar and Brown1979, Ramirez Verdugo & Romero Trillo Reference Ramirez Verdugo and Romero Trillo2005), examining the intonation of tag questions can be accomplished only once the meaningful tunes – for this particular variety from northwest England – are catalogued. Cruttenden Reference Cruttenden2001 contends that the intonational system of a very closely related variety, the Mancunian dialect, presents a particular problem for standard systems of intonational representation. Nonetheless, we recognize the value of an intonational analysis and plan to pursue this line of research in future work.

Similarities Across Groups

We begin discussing our results by identifying properties of tag questions that are similar or identical across the four communities of practice, and we conclude the section by discussing the social significance of such similarities. In spite of their widely divergent social styles, the four groups use tag questions in some discursively similar ways. Perhaps unsurprisingly, all four groups use tags at the ends of their turns the majority of the time. As indicated in Figure 2, a greater percentage of tag questions appear turn-finally than turn-medially (χ² = 25.918, df = 1, p ≤ 0.001), with no significant difference between social groups.

FIGURE 2:

Location of tag relative to speaker turn

In addition to this pattern of tag production, we find cross-group similarity in how interlocutors respond to turn-final tags. As shown in Figure 3, for all four groups, interlocutors most often express explicit agreement with turn-final tags (χ² = 222.583, df = 2, p ≤ 0.001), as opposed to expressing dissent or offering no response, with no statistically significant differences between the four groups.

FIGURE 3:

Agreement with turn-final tags

Having established that tag questions most commonly occur at the ends of turns and that interlocutors most commonly agree with their propositional content, regardless of social group, we now turn to consider the meaning of these patterns. First, the fact that most tags are turn-final suggests that they are frequently used by speakers prior to conceding the conversational floor. That is, they are used to encourage orderly and shared turn-taking. This is consistent with Sacks, Schegloff & Jefferson’s (Reference Sacks, Schegloff and Jefferson1974:719) characterization of tag questions as exit devices. Furthermore, the fact that they most frequently receive agreement – and rarely dissent – suggests that they encourage recipients to take a stance in response, specifically one of alignment (Du Bois Reference Du Bois and Englebretson2007). This interpretation concurs with Hudson’s (Reference Hudson1975) analysis of the syntax and semantics of tag questions. He observes that tag questions carry the sincerity conditions of both a declarative and a question: much as with a declarative, “the speaker believes the proposition to be true,” and as with an interrogative, the speaker of a tag “believes that the hearer knows at least as well as he does whether or not the proposition is true” (Hudson Reference Hudson1975:24). Kimps (Reference Kimps2007:272) similarly notes that “tag questions convey the speaker’s orientation to the proposition by signaling a specific attitude and the expected response.” Tags are conducive, then, because they encourage the hearer to agree with a proposition – in the sense that it is easier to agree with an established proposition than to construct dissent against it – particularly when an interlocutor is responding to the first part of an adjacency pair. This is borne out in the quantitative results for the discourse shape of tags. These observations suggest that tag questions may have a basic function that is universal across social groups: being conducive. We use the term “conducive” in contrast to “facilitative.” The term “facilitative,” as it is commonly employed in the language and gender literature, implies altruism and positive politeness on the part of the speaker. In contrast, a speaker can be conducive for a number of different reasons – including being manipulative or enlisting support for a contentious claim. Only qualitative analysis could expose the nature of the conducing in the case of each individual tag question; we return to this issue below, in discussing the ways in which members of each CofP distinctively style tags.

The conducive nature of tag questions provides us with a starting point to begin elaborating on their social meaning. Recall that earlier we noted how Eckert Reference Eckert2008 places the myriad social meanings that could be attributed to a linguistic feature in an indexical field. If a variable has an underlying function that goes across its use, we may consider this core association, or “kernel of similarity” (Podesva Reference Podesva2008), to lie at the center of the indexical field as an n-th order indexical. In the case of tag questions, this n-th order indexical is “conducive.” Of course, the indexical field of tag questions extends well beyond the realm of the n-th order indexical value we have identified. We begin to move into the territory of higher-order indexical meanings in the following section, in which we examine differences in tag question usage between the four CofPs.

Distinctions Between Groups

In this section, we focus on stylistic practices at the discursive, grammatical, and phonological levels of tag question design. As discussed in the introductory section, we follow Irvine Reference Irvine, Eckert and Rickford2001 in treating style as distinctiveness and aim primarily to identify the differences between groups that emerge through quantitative analysis of the data. The social significance of these distinctions will be discussed in the next section. We begin by discussing intergroup differences in discourse properties of tag questions.

Recall from the previous section that speakers of all groups typically expressed agreement with turn-final tags. If we consider the rate of agreement with turn-medial tags, however, we find a different pattern. As is evident in Figure 4, most groups – the Populars, Townies, and Geeks – usually offer no response to turn-medial tags. Put another way, members of the Popular (χ² = 13.562, df = 1, p ≤ 0.01), Townie (χ² = 4.184, df = 1, p ≤ 0.06),Footnote ⁹ and Geek (χ² = 18.157, df = 1, p ≤ 0.01) CofPs express agreement more often in response to turn-final tags than to turn-medial tags. We argue that this pattern obtains because the pragmatic pressure to concur with the proposition expressed in a tag question is reduced when the tagger does not concede the conversational floor.

FIGURE 4:

Agreement with turn-medial tags

Shifting our attention to the Eden Village girls, the data reveal that they typically express agreement even with turn-medial tags. Though interactional norms do not necessitate the expression of agreement turn-medially, Eden Village girls provide it nonetheless. One can interpret this pattern to mean that Eden Village girls are particularly agreeable, or that they are reluctant not to respond to tags (Figures 3 and 4 both illustrate that of the four CofPs, Eden Village girls have the lowest percentage of “no response” responses). Either interpretation supports the claim that these girls exhibit a high-involvement style (Tannen Reference Tannen2005 [1984]). We find additional evidence for this claim if we examine the interaction between the expression of agreement and overlapping. Figure 5, which depicts the location of overlap relative to tags that were agreed with, shows that most tags are delivered turn-finally without overlap. More relevant to the point at hand, the figure indicates that Eden Village girls resist overlapping less and more frequently overlap with the tagger during the tag itself. Eden Village girls, then, exhibit a strong tendency for what Tannen Reference Tannen1990 terms “cooperative overlap,” which we operationalize as the co-occurrence of overlap and agreement.

FIGURE 5:

Location of overlap relative to tag when expressing agreement

Shifting gears from variation above the level of the utterance to variation within the utterance, we focus our attention on the grammatical subject of the tag question. Unsurprisingly, we found that the grammatical subject used interacts with the content of the talk. For example, first person subjects are relatively common when groups are talking about their own CofP, but considerably less so when discussing other CofPs. Perhaps more interesting are the correlations between grammatical subject and CofP membership when individuals are not talking in the first person. When talking about their own group, Townies use relatively higher rates of he subjects in tags (14% compared to between 4–8% for the other CofPs). It is noteworthy that other groups generally did not talk about boys in contexts where they produced tags. All other groups, when talking about themselves, have high rates of she subjects in tags (Townies = 16%; other CofPs between 23–28%), indicating that they talked about friends in their own groups above anything else. The Populars and Eden Village girls also use high rates of we (c.25%, compared to c.13% for the Townies and the Geeks), suggesting that tags occur in collective and self-inclusive styles of talk. When talking about other groups, Townies’ tag questions have a relatively high number of they subjects (40%, compared to between 13–28% for other CofPs), while other groups tend to produce tags with more she subjects (7% for Townies, compared to between 28–63% for other CofPs). When talking about generic topics, all groups unsurprisingly prefer it subjects (and recall that invariant innit is not represented in the analysis). This finding is consistent with Tottie & Hoffmann’s (Reference Tottie and Hoffmann2006) observation that it is the most common pronoun in British English tag questions (they also exclude invariant innit from analysis), usually in the isn’t it? form. In analyzing the grammatical subject of the tag questions, we are given an indication of the wider discourse context of the tags’ occurrence. Given that tags are attributed social meaning on the basis of their contextual use, we discuss how the content of talk affects how a tag may be construed in the next section.

With regard to morphosyntax, we found that all groups predominantly produce standard morphosyntax, but the extent to which they do this correlates with their placement on a pro-/anti-school continuum (see above). As shown in Figure 6, the association between nonstandard grammar use and CofP is statistically significant (χ² = 27.634, df = 3, p ≤ 0.01), with Townies producing far more fully lexical nonstandard forms than the other groups. In related work, Moore Reference Moore2003 investigated the effect of CofP as well as a number of internal constraints (verb function, subject type, and polarity) on the rate of nonstandard were, the most commonly used nonstandard morphosyntactic form. She reports that CofP strongly constrains its occurrence.

FIGURE 6:

Nonstandard morphosyntax

The last difference in tags between groups concerns a phonological feature, the realization of word-final /t/. Making a three-way distinction between deleted, glottalized, and released stops, we find that the distribution of variants varies significantly according to CofP membership (χ² = 26.089, df = 6, p ≤ 0.01). As shown in Figure 7, all groups use the glottalized form most frequently and do so with the same frequency. While previous work observes extensive social differentiation in various degrees of glottalization (e.g., Docherty et al. Reference Docherty, Foulkes, Milroy, Milroy and Walshaw1997, Llamas Reference Llamas2007, Milroy et al. Reference Milroy, Milroy, Hartley and Walshaw1995), a multivariate analysis did not choose CofP as a factor influencing the use of this feature. We attribute this result to the fact that social class, gender, age, and geographical region – none of which varies significantly among speakers represented in the current study – are the most robust social dimensions of variation for this particular feature.Footnote ¹⁰ With regard to the other variants, Geeks release /t/ more than members of the other communities of practice, a finding which has also been reported by Bucholtz Reference Bucholtz and Ahlers1996, Reference Bucholtz2001 in her work on the linguistic practices of “nerds.” By contrast, Townies exhibit high rates of deletion when compared to the other three CofPs. Both of these differences obtain in VARBRUL analyses that factor in the effect of linguistic factors influencing the choice of one variant over another.

FIGURE 7:

Realization of word-final /t/

Table 1 summarizes the factors influencing the occurrence of released /t/. The following phonological environment is the strongest predictor of the released variant, with vowels strongly favoring it and consonants strongly disfavoring it. While previous work on linguistic factors affecting the rate of released /t/ (Benor Reference Benor2001, Podesva Reference Podesva2006) reports the significance of the preceding phonological environment and morphological affiliation, both of these factors are invariant in the data analyzed here, as all tokens are preceded by /n/, from the contracted form of not (e.g. isn’t). When taking the following environment into account, CofP still has a significant effect, with Geeks highly favoring the feature, Populars slightly favoring it, and Townies and Eden Village girls avoiding it nearly categorically.Footnote ¹¹

TABLE 1.

VARBRUL summary of factors influencing occurrence of released (-t) in tag questions.

Input = 0.012, Log likelihood = -92.018, p ≤ 0.000

Table 2 encapsulates the results of a multivariate analysis used to determine factors influencing the occurrence of the deleted variant of word-final /t/. As was the case for the released variant, the following environment has the strongest effect on the occurrence of deletion, with vowels strongly encouraging deletion and consonants strongly discouraging it. This trend runs counter to every study on (-t/d) deletion, whether on White (Guy Reference Guy and Labov1980, Reference Guy1991), African American (Fasold Reference Fasold1972, Labov, Cohen, Robins & Lewis Reference Labov, Cohen, Robins and Lewis1968, Wolfram Reference Wolfram1969), or Chicano (Santa Ana Reference Santa Ana1996) varieties of U.S. English or (White varieties of) British English (Tagliamonte & Temple Reference Tagliamonte and Temple2005). We attribute this finding to the phonotactics of /t/ within tag questions. As mentioned previously, all of our (-t) tokens are preceded by /n/, whereas previous studies have examined the full range of preceding segments. In the environment n __ # C, /t/ is most commonly realized as a glottal stop or glottalized stop. On the other hand, in the environment n __ # V, /t/ is most frequently deleted, since the glottal variant is less permissible here. Regarding social constraints on the use of the deleted variant, Townies slightly favor the deleted form, while the other groups slightly disfavor it.

TABLE 2.

VARBRUL summary of factors influencing occurrence of deleted (-t) in tag questions.

Input = 0.239, Log likelihood = -227.903, p ≤ 0.047

To summarize the results of this section, we find a number of distinctive properties of each group’s tag questions. Compared to members of the other CofPs, Populars use relatively high rates of she and we as grammatical subjects when talking about themselves. In contrast, Townies are more likely to use he as a grammatical subject in the same context. They also use more vernacular grammatical and phonological forms, exhibiting higher rates of nonstandard verb morphosyntax and deleted /t/. Geeks use more conservative phonology in tag questions, using relatively high rates of released /t/. Finally, Eden Village girls, like the Populars, more frequently use she and we as grammatical subjects when discussing their own CofP. They differ from the Populars and the other CofPs, however, mostly with respect to the discourse design of tag questions, very often expressing agreement as interlocutors (even when speakers are in mid-turn) and often overlapping during the tag itself to express agreement. In the following section we look more closely into each CofP’s distinct practices of tag question usage, locating tag questions in their ethnographic and discourse contexts in order to tap into the range of social meanings they can index.

Tagging Styles

In this section, we focus on how members of the four communities of practice diverge in their use of tag questions and how divergences assist members in constructing distinct group styles. The girls whose stylistic practices are under analysis in this article combine features of discourse, grammar, and phonology in ways that contrast with those of their peers at Midlan High. This section is concerned with examining how constellations of each group’s distinct features of tag questions are employed in interaction and with situating these linguistic practices in relation to social practice. We show that even though all groups use tag questions, and that even though all of them use tags to conduce, what precisely is conduced varies greatly from one group of girls to the next.

For each of the four groups, we will discuss conversational extracts featuring the group’s prototypical tag design. That is, we examine excerpts featuring all or many of the tag components that make the group linguistically distinct from its peer groups, as identified in the quantitative analysis discussed in the previous section. It is important to note that we do not intend to suggest that every tag uttered by every CofP member is designed like the examples discussed. Rather, the examples are presented as illustrations of the group’s prototypical tag questions, which we take to represent iconic performances of group style. These constellations of features, or style clusters, “provide further insight into the social motivations for employing styles” (Podesva Reference Podesva2008:7).

The Popular tag

The extract in (1) features a clustering of eight prototypically Popular tag questions, each appearing in boldface. In this excerpt and all that follow, we use [ ] to indicate overlapping speech, < > for descriptions of nonlinguistic sounds or gestures, .. for pauses, and – for hesitations or disfluencies. The phonetic realization of /t/ in tag questions is represented in IPA, enclosed by square brackets; deleted variants are represented by [ø]. The girls in this interaction are talking about the substantial discord between Ellie, a Townie girl, and Annabel, another Popular girl.

1. (1) 1 Cindy: And I think.. at one point – at – near the end of Year 9, they changed for the

   2 worse, some of them, din[ø] they? [They were really] awful. But..

   3 Tina: [Yeah, cos like.. yeah]

   4 Cindy: they’ve calmed down a bit now, [but..]

   5 EM: [What,] like being bitchy about

   6 [people and stuff?]

   7 Tina: [Yeah, cos people –] some people like noticed more than others. D’you know

   8 what I mean?

   9 Cindy: Um.

   10 Tina: Like.. especially like Annabel.

   11 She – cos Ellie and Annabel don’t really get along, do they?

   12 Cindy: No.

   13 Tina: They’re only – they’re only really civil to each other. And they only get along if

   14 like they’re in really good – if they’re in a really good mood. Say, if one of them’s

   15 just in a bad mood – it’s s-, nothing to do with the other one – they’ll take it out

   16 on that one.. on the other person.

   17 Cindy: They don’t get on at all.

   18 Tina: No.. They could be fine and then if someone brings summat up, one of them’ll

   19 say something about her. D’ [you know what I mean?]

   20 Cindy: [Yeah.]

   21 Tina: D’you [know what I mean? So..]

   22 EM: [I noticed when I came in like doing this – ] like yesterday, as soon as

   23 Ellie came in the room, Annabel went out.

   24 Tina: Yeah.

   25 EM: They just won’t talk, will they?

   26 Tina: No. They just – they’ve never got along really, have they? Cos when they were at

   27 primary school, they – that’s what we’ve heard anyway.

   28 Cindy: Yeah.

   29 Tina: So..

   30 EM: So are you still getting along with Annabel and stuff?

   31 Tina: Well, yeah. We don’t despise her but she’s – she can be a bit annoying. She’s there

   32 going, “<Sharp intake of breath>.” [Hyperactive,] in[ø] she?

   33 Cindy: [Loud.] Well, cos me and Annabel were

   34 quite close, but we’re not really as much any-, anymore, are we?

   35 Tina: No.

   36 Cindy: As much as we used to be. [Which] is quite good

   37 EM: [Oh.]

   38 Cindy: [<laughs>]. Cos I think it –

   39 EM: [<laughs>]

   40 Cindy: umm, we were too like together and it just..

   41 Tina: Yeah, it kind of annoyed you a bit, [dint[t^h] it, when she was] babbling away?

   42 Cindy: [Yeah, I dint like it.] Controlling and..

   43 Tina: Yeah.

   44 Cindy: And I just dint like it.

   45 Tina: She’d be like really nasty to you [about stuff, weren’t[ʔ] she?]

   46 Cindy: [Yeah. Being really awful.]

   47 Tina: Like if she laughed or summat, she’d scream out,

   48 “Cindy, shut the fuck up!”

   49 Cindy: Yeah, if I laughed all the time.

   50 Tina: Cos Cindy’s like really always laughing.

   51 Cindy: <laughs>

   52 Tina: Cos me and Cindy really get along. We just laugh about [anything.]

   53 Cindy: [Cos we’re] the only

   54 ones that get – get each other’s jokes, aren’t[ʔ] [we? Nobody gets our] jokes.

   55 EM: [<laughs>]

Recall that, in our analysis of grammatical subject, we noted that grammatical subject interacted with the content of the talk in expected ways (e.g., more first person subjects in talk about one’s own CofP). When exploring this interaction, we noted another stylistic peculiarity: Different CofPs tended to produce their tags in slightly different kinds of talk. As Figure 8 illustrates, most tags occur in talk about one’s own CofP, irrespective of the CofP membership of the tag producer. However, Populars produce proportionally more of their tags in talk about other groups than do other CofPs; whereas Townies produce proportionally fewer tags in this context. Note also that Geeks produce proportionally more of their tags when talking generically (i.e., in talk that does not relate to any of the speaker’s high school peers or extended social group) than do other CofPs. The association between CofP and the type of talk in which tags occur was found to be statistically significant (χ² = 36.712, df = 6, p ≤ 0.01).

FIGURE 8:

Type of talk in which tag questions occur

What distinguishes the tag questions in (1) as Popular tags, then, is their occurrence in talk about a girl belonging to another group (lines 1–2, 11, 26), and the tendency to use more she (lines 32, 45) and we (lines 33–34, 53–54) subjects in tags occurring in talk about their own group. An analysis of what Cindy and Tina accomplish in this extract provides us with a story about what this particular tag question design achieves. The extract shows that for Populars, tags occur in talk about people; more specifically, they are used to describe, evaluate, and position Populars individually and collectively relative to other people in their school. Tina contends in line 11 that Ellie and Annabel – one a Townie, the other a Popular – do not get along, attempting to enlist Cindy’s support in constructing opposition by asking a tag question. Cindy agrees with Tina’s assertion, explicitly assenting in line 12. Tag questions, as used by the Populars, seem to occur in talk that could be construed as gossip, which is often said to describe and evaluate (Coates Reference Coates and Coates1998, Gluckman Reference Gluckman1963, Jones Reference Jones and Cameron1990). For the Popular girls, tag questions – occurring as they do in talk that aims to conduce agreement around social engineering, identities, and self-presentation – may be an effective linguistic strategy in gossip contexts because of their ability to emphasize or intensify criticism. Later in the extract, after the opposition between Ellie and Annabel has been fully established, Tina and Cindy police norms within their own group. They claim Annabel as a member of their group (line 31), but not without a strong dose of criticism for hyperactivity (line 32), talking too much (line 41), and acting nasty (line 45). Tina and Cindy use tag questions in the context of criticism to jointly conduce a critical view of behaviors they deem unacceptable, which they contrast with more desirable behaviors, such as getting each others’ jokes (lines 53–54). In the latter case, a tag question is used to conduce agreement over what is acceptable.

The Populars themselves acknowledge the regulatory function that their tags serve. In post-fieldwork interviews, Moore asked a few Populars about their use of tag questions and whether they use particular variants. As Beverley observes in line 4 of the extract in (2), Populars utilize tag questions as a means of criticizing, with criticism generally falling on another girl, as suggested by the tag’s pronoun shift from line 1 to line 4.

1. (2) 1 EM: “In[ø] ‘e?”

   2 Beverley: We do that.

   3 Lindsey: Yeah.

   4 Beverley: When you – when you’re slagging someone off. “Int[ʔ] she?” Yeah <laughs>.

In addition to the distinct functions Popular tag questions serve, it is important to note that Populars differ from members of the other CofPs in the frequency with which they use tags. A frequency analysis for tag questions cannot follow Labov’s (Reference Labov1969) principle of accountability, as it is impossible to identify the number of potential environments in which tags could appear, a well-documented challenge of studying variation in discourse features (Lavandera Reference Lavandera1978, Romaine Reference Romaine1984). Accordingly, we did not incorporate the frequency dimension into the study detailed in the third section above, preferring to follow the more established tradition of reporting the number of tags occurring in a given number of words as employed, for example, in sociolinguistics (Macaulay Reference Macaulay, Chambers, Trudgill and Schilling-Estes2002) as well as corpus linguistics (Tottie & Hoffmann Reference Tottie and Hoffmann2006). As shown in Figure 9, Populars use considerably more tag questions than members of the other three groups. While a one-way between groups analysis of variance showed no statistically significant difference in tag frequencies for the four groups, a planned contrast test indicated a statistically significant difference between the scores of the Populars and those of the members of the other three CofPs (F(1, 23) = 5.09, p ≤ 0.034). The tagging rates for the Townies, Geeks, and Eden Village girls is comparable to the rate reported in Tottie & Hoffmann (2.38 tags per 1,000 words) for British English, based on the spoken component of the British National Corpus. The Popular rate, in contrast, stands considerably higher, at 4.37 tags per 1,000 words. Moreover, there is considerable intragroup homogeneity in tagging rate; when the 27 girls whose speech is represented in the corpus are ranked according to individual tagging frequency, the four highest – each with a tagging rate above 6 tags per 1,000 words – all belong to the Popular CofP.

FIGURE 9:

Mean frequency of tag use

With the tagging rate for Populars so much higher than that of the other groups, it is possible that the act of tagging may be indexically linked to membership in the Popular group itself. Indeed, Populars are recognized and recognize themselves as active users of tag questions. In Moore’s post-fieldwork interviews, for example, the Popular girls consulted acknowledged that using tag questions was a behavior in which they habitually engaged, as exemplified above in extract (2).

We argue that the association between tag questions and Popular status began to surface in Year 9. As mentioned in the third section, it was at the end of this year that the issue of group identity took center stage, when the Populars and Townies split into two groups. This occurrence, which saw the Townies acknowledged as the most rebellious CofP, left the Populars’ identity in question, given that they had previously occupied this role. While the Townies’ nonstandard morphosyntax assisted in their maintenance of a “cool” identity, through the mediating stance of rebelliousness, Populars did not use nonstandard features to the same extent. As shown in the preceding section, Popular tag questions are only moderately vernacular in terms of morphosyntactic design. Instead, using tag questions – particularly in the way exemplified by the style cluster appearing in (1) – served as a resource for Populars to construct themselves as simultaneously cool (as a group whose membership was worth monitoring) and distinct. Moore Reference Moore2004 provides evidence for this assertion, reporting that the frequency with which Populars used tag questions increased from Year 9 to Year 10. This trend indicates that tags began occupying a more significant part of the group’s linguistic repertoire.

The Indexical field for Tag Questions

Having examined intergroup distinctiveness in tag design in the previous section, we are now in a position to elaborate on the indexical relationships between tag questions and their many social meanings. Returning to the concept of the indexical field (Eckert Reference Eckert2008), as introduced in the first section, we note that the indexical field relates the direct index of a particular linguistic form to higher-order indexical values. Put in Silverstein’s (Reference Silverstein2003) terms, the indexical field represents the ideological associations among an n-th order indexical value and any additional (e.g. n + 1st order) indexical values. As Eckert (Reference Eckert2008:484) asserts, “The emergence of an n+1st indexical value is the result of an ideological move, a sidestepping within an ideological field.” In other words, creating an n + 1st order indexical value simultaneously reconstrues the n-th order indexical value and gives shape to the linguistic feature’s indexical field. Thus, at the heart of the indexical field for tag questions lies the n-th order indexical value we identified, and radiating out from it is a series of n + 1st order indexical values. This relationship is captured in Figure 10.

FIGURE 10:

Indexical field for tag questions at Midlan High

Recall that the direct index for tag questions was “conducive,” depicted in the box at the center of the figure. Conduciveness is enlisted in constructing many other kinds of social meanings (n + 1st order indexical values), as was discussed in detail in the preceding section. These n + 1st order indexes appear in plain text in Figure 10. Populars use tag questions to conduce evaluative stances toward girls in the school, and to criticize and regulate behavior. Such uses of tag questions have enabled Popular girls to portray themselves as cool. In contrast to Populars, Townies use tag questions to conduce a shared viewpoint surrounding their independence, experience, and authoritativeness. But similar to the Populars, the Townie-specific use of tags can also index their own brand of coolness. Thus the “cool” second-order index is recruited by both Populars and Townies, illustrating that meanings in the indexical field can be repackaged and combined in unique ways to create quite distinct local identities. The Geeks also draw on the conducive functions of tags to construct authority and emphasize their knowledge. Here again we see overlap in the meanings exploited by two different groups, as both Geeks and Townies draw on the “authoritative” n+1st order indexical value. Finally, Eden Villagers are intensely collaborative in their use of tags, politely conducing their mutual involvement and friendships. Like the Populars, Eden Villagers utilize the regulatory functions tags cans serve.

We have until now focused mainly on social meaning at the interactional level (mostly stances and ephemeral characteristics of individuals at particular moments in time), but these meanings are, of course, not divorced from more enduring socially meaningful units like personae and social types. Johnstone and her colleagues have shown, for example, that the indexical values of a number of linguistic features can be harnessed for indexing identities tied to the local geographic area of Pittsburgh. Johnstone Reference Johnstone and Englebretson2009, for example, illustrates how the discursive construction of an expert stance, built through the demonstration of dialect awareness, enables an individual to construct a Pittsburgher identity. Phonological features can be used to similar effect, as monophthongal /aw/ has over time been linked to both working-class identity and Pittsburgh (geographical) identity (Johnstone et al. Reference Johnstone, Andrus and Danielson2006, Johnstone & Kiesling Reference Johnstone and Kiesling2008).

In the current study, the indexical values of tag questions can come to index a group persona through the process of stance accretion (Du Bois Reference Du Bois2002; Rauniomaa Reference Rauniomaa2003, both cited in Bucholtz & Hall Reference Bucholtz and Hall2005). For example, we have seen how Populars use tag questions in order to construct cool, critical, and evaluative stances. The frequency of the Populars’ repeated co-production of these stances – revealed in Figure 9 and acknowledged in (2) – hints at an ideological crystallization, to the point that the stances accrete into the Popular categorization itself as a higher-level indexical order (represented in small caps in Figure 10). A particular tag design, and its component indexical meanings, may thus become associated with a CofP identity. Apart from the Populars, it is unclear whether the entire set of stances associated with each group has accreted into a discrete social meaning emblematic of a group; accordingly, the CofP labels in Figure 10 lie outside the borders of the indexical field, and the boundaries between CofP styles are marked with dotted lines.

Finally, associations between linguistic features and identities do not end at the locally meaningful meso-social level. All the CofPs we have analyzed in this article are also constrained by and judged relative to macro-category classifications. For instance, the Townie girls’ rebellious lifestyle causes them to engage in areas of the wider community that are more readily associated with working-class culture; so despite the relatively high social status of these girls, their forms of practice and engagement regularly situate them (and their tag question use) relative to a working-class social type. Furthermore, while all of the girls analyzed in this study unambiguously classify themselves as female and are identified by others as such, the nature of some girls’ practices sit more comfortably alongside dominant discourses of femininity. Where this is the case, for instance, with the Eden Village girls (who by their own admission are “girly girls”), tag questions may be interpreted as part of this overall “feminine” style. Indexical values at the macro-social level are represented in capitalized italics in Figure 10.

Of course, the Eden Village girls and their practices can be interpreted as “feminine” and the Townies and theirs can be interpreted as “working-class” only if the interpreter engages with an ideology that facilitates such a link. The same can be said for all the indexical associations we have made in this section. Because we have deconstructed the trajectory of meaning in this community, we are able to trace social meanings from a core “conducive” meaning, via the micro-social indexical level of stance, through the stylistically embedded meso-social level of personae and macro-social level of demographic social type. In this way, we have “layered” meaning across the identity frames acknowledged by Bucholtz & Hall Reference Bucholtz and Hall2005 and Coupland Reference Coupland2007 and provided the analyst’s view of social meaning at Midlan High. However, it is important to note that, at any one time, for the Midlan High speaker these meanings may be discrete and in competition with one another; thus, the same tag question uttered by a Popular girl may be variably heard as “Popular” or “cool” or “evaluative” depending upon who at Midlan High hears it and the ideological lens through which it is filtered. One of the benefits of the indexical field we have presented is that it enables us to see meanings as distinct (albeit related) and, thus, discourages an analyst from privileging any one level of meaning without appropriate contextual accountability.