Reconsidering the Social in Language Learning: A State of the Science and an Agenda for Future Research in Variationist SLA

Abstract

The current paper offers a critical reflection on the role of the social dimension of the second language (L2) development of sociolinguistic competence. We center our discussion of L2 sociolinguistic competence on variationist approaches to second language acquisition (SLA) and the study of variable structures. We first introduce the framework of variationist SLA and offer a brief overview of some of the social, and more broadly extralinguistic, factors that have been investigated in this line of inquiry. We then discuss the three waves of variationist sociolinguistics and various social factors that have been examined in other socially oriented approaches to SLA. By reflecting on these bodies of research, our goal is to identify how the insights from this work (i.e., research couched in the second and third waves of variationist sociolinguistics and in other socially oriented approaches to SLA) could be extended to the study of L2 sociolinguistic competence. We argue that greater attention to the social nature of language in variationist SLA is needed in order to more fully understand the L2 development of variable structures.

1. Introduction

The current paper aims to contribute to scholarship on sociolinguistic competence (see Kanwit, 2022, and the other articles in this Special Issue) by offering a state of the science of and an agenda for future research on the social, and more broadly the extralinguistic, dimension of language that shapes second language (L2) development and usage.1 Sociolinguistic competence refers to the ability to use and interpret the social meaning of language (Geeslin & Long, 2014). We couch our reflections within variationist approaches to second language acquisition (SLA; Bayley et al., 2022; Bayley & Tarone, 2012; Geeslin & Long, 2014). Central to traditional variationist SLA is the recognition that language varies according to numerous factors, not only linguistic but also extralinguistic (i.e., social, individual, contextual).2 Variationist SLA is an important strand of research in applied linguistics because it has made valuable contributions to the understanding of learners’ acquisition of sociolinguistic variation. Despite these insights, we focus this state-of-the-science article on an issue in need of more rigorous inquiry than it has received in previous research: the social side of language variation. We aim, therefore, to initiate a call for a more concentrated focus on social variation in order to develop a stronger theoretical understanding of the ways in which such variation shapes L2 development.

We begin by briefly reviewing why it is important to study the extralinguistic side of L2 acquisition. We then turn to variationist SLA and examine the extralinguistic factors that have received the most attention in this area of research. These include gender, social class, learning context, L2 proficiency, and discourse topic (e.g., Adamson & Regan, 1991; Geeslin, 2011, 2020; Kanwit & Solon, 2013; Mougeon et al., 2010; Regan et al., 2009). Next, we consider the ways in which variationist sociolinguistics has evolved over its three waves (Eckert, 2012) and how these developments could be valuable for L2 variationism. We then discuss extralinguistic characteristics that have been investigated in other socially oriented approaches to SLA. Examples of these factors include learner sexuality and gender identity (e.g., Knisely, 2020), race and ethnicity (e.g., Alim et al., 2016; Flores & Rosa, 2019), and social distance between interlocutors (e.g., Félix-Brasdefer & Shively, 2021). This review identifies extralinguistic characteristics shown to be important in L2 acquisition that have not been fully integrated into work in variationist SLA. Throughout our discussions of the three waves of variationist sociolinguistics and extralinguistic considerations in other approaches to SLA, we offer a critical reflection on the need to investigate extralinguistic variation in a more profound way than we currently do in most L2 variationist research. We argue that more fully incorporating the extralinguistic dimension of language into studies of variationist SLA is vital not only for advancing theoretical predictions about the acquisition of variation but also for furthering knowledge about sociolinguistic competence.

2. Variationist SLA

Since SLA’s inception, cognitivism has guided much of its scope of investigation (Atkinson, 2011); this disciplinary orientation has meant that traditionally the field has sought to understand how an L2 is acquired in the mind. Notwithstanding the value of this scholarship, over the years, researchers have advocated for an expansion of the field to further account for social facets of language (e.g., see early works in variationist SLA, such as Bayley, 1991; Dickerson, 1974; Preston, 1993).3 Subsequently, Firth and Wagner (1997), for example, argued that “language is not only a cognitive phenomenon, the product of the individual’s brain; it is also fundamentally a social phenomenon, acquired and used interactively, in a variety of contexts for myriad practical purposes” (p. 296). In short, investigating the social, or extralinguistic, dimension of L2 acquisition is valuable because language is inherently social: it is used in social contexts to express social meanings reflective of the identities of the interlocutors (Eckert, 2000). A discussion of the debate about SLA’s scope of inquiry (i.e., cognitive versus social) is beyond the purview of the article, so we simply note that SLA has diversified its scholarly focus and consequently its theoretical orientations (Atkinson, 2011). Importantly, there now exists evidence of the complex ways in which the extralinguistic dimension of language shapes L2 acquisition (see Tarone, 2000; Geeslin & Long, 2014).

2.1. An Overview of Variationist SLA

One social approach to acquisition is variationist SLA, which seeks to explain variation in learner language, namely, the complex ways in which learners’ language usage is variable and in which variation/variability is important for understanding acquisition. In this area of research, variation refers to the use of two or more forms (i.e., variants) to express the same meaning or function. A phonetic example in Arabic is variation with the character ج jīm (Raish, 2015). Various realizations exist, with [ɡ] being associated with Egyptian dialects and [dʒ] with Levantine varieties. Verbs that express futurity in French (e.g., periphrastic future, inflectional future, and present indicative) constitute an instance of morphosyntactic variation (e.g., Gudmestad et al., 2020), as shown below.

• Periphrastic future: Demain je vais aller à Londres. “Tomorrow I am going to go to London.”
• Inflectional future: Demain j’irai à Londres. “Tomorrow I will go to London.”
• Present indicative: Demain je vais à Londres. “Tomorrow I go to London.”

Central to variationist SLA, which has roots in variationist sociolinguistics (e.g., Labov, 1972), is the recognition that language varies according to numerous factors, both internal and external to the language system, where internal factors are linguistic features of the language itself and factors external to language include social, individual, and contextual characteristics of speakers and the communicative context.4 We continue with the aforementioned examples in Arabic and French to offer examples of linguistic and extralinguistic factors. First, Raish (2015) investigated variation in the phonetic realizations of ج jīm among L2 and (Levantine and Egyptian) heritage learners of Arabic who were studying in Cairo, Egypt. The linguistic factor of word position and the social factor of gender were two constraints that impacted participants’ use of ج jīm. The Egyptian variant [ɡ] was favored in word-initial position and among women. In the study on future-time expression and French, lexical temporal indicator and topic seriousness5 were internal and external factors, respectively, found to impact the use of future-time verbs in Gudmestad et al. (2020). Near-native speakers of French living in France were more likely to use present indicative to express futurity when a time-specific lexical temporal indicator (e.g., aujourd’hui “today”) was present in the same clause as the future-time verb, and they were less likely to use present indicative when the topic of conversation was serious (e.g., education).

Variationist SLA distinguishes between two types of variation: Type I or vertical variation and Type II or horizontal variation (Rehner, 2002). Type II variation has been the focus of most variationist SLA research. It pertains to linguistic phenomena that are variable among native speakers.6 In sociolinguistics, these are called linguistic variables, and in SLA, they are termed variable structures. We use the latter term in the current paper. Phonetic variation with the Arabic character ج jīm and future-time expression in French are examples of variable structures. Instances of Type II variation offer a unique acquisitional challenge to learners: Because native-speaker usage is variable, the input learners receive is presumed to be characterized by this variability and the target of acquisition is also variable. Categorical usage, in other words, is typically not the learning goal.7 This is in contrast to Type I variation, where variability is not (generally) seen among native speakers but is present among learners as part of their language development. For example, in Portuguese, certain prepositions (e.g., a “to”, em “in”, Picoral & Carvalho, 2020) categorically contract with the following definite article (e.g., em + a = na “in the_FEM.SG”). Picoral and Carvalho (2020) investigated the variable use of these contractions (i.e., use versus non-use) among third-language learners of Portuguese in order to account for the Type I variation seen in this learner population.

The psycholinguistic model of interlanguage variation, developed by Dennis Preston (2000, inter alia), offers a theoretical framework for explaining variability in learner language. This model includes three explanatory levels. Level 1 consists of social (or extralinguistic) variables, including the aforementioned factors like gender of the speaker, L2 proficiency, and topic seriousness. Level 2 pertains to linguistic factors. Word position and lexical temporal indicators described above for ج jīm and future-time expression, respectively, are examples of linguistic features that can constrain variable usage. Finally, Level 3 is time, which accounts for the variability observed in learner language along the developmental trajectory. Learners’ use of a linguistic form in this model is explained in probabilistic terms, with a particular variant more or less likely depending on the specific combination of linguistic and social constraints that characterize a given instance of usage and by the point at which learners are on the developmental trajectory (i.e., time). This model, thus, offers a way of conceptualizing and empirically investigating the complex interplay between the linguistic and social dimensions of language in L2 acquisition (Tarone, 2000).

The methodological tools that variationist SLA researchers employ align with Preston’s (2000, inter alia) theoretical account. The analyses tend to be quantitative and consist of two general components. First, researchers document the frequency with which each variant occurs in the dataset. Second, they conduct a multivariate analysis (usually a regression model) to explain learners’ variable behavior. In a regression model, the linguistic phenomenon under investigation is the dependent variable (e.g., phonetic realizations of ج jīm in Arabic, future-time expression in French, preposition + article contractions in Portuguese). The independent variables are the internal and external factors thought to condition variation. The model reveals which factors are important and the specific role they play in variation. For example, Picoral and Carvalho (2020) found that one of the significant linguistic constraints on preposition + article contraction was article number: learners were more likely to use the contraction when the article was singular compared to plural. A significant extralinguistic factor was third-language Portuguese level: Level 3 (the most advanced group) favored contraction, Level 2 disfavored contraction, and Level 1 neither favored nor disfavored contraction. In this way, a regression analysis has the ability to explain how multiple linguistic and social factors simultaneously influence variability, rather than accounting for factors in isolation.

Variationist SLA is a valuable area of research in applied linguistics because it helps to further the understanding of how learners build a more nuanced grammar that reveals sensitivity to linguistic and extralinguistic factors. It also provides a model for integrating the extralinguistic dimension of language into theoretical understandings of L2 acquisition. Despite these important contributions, we believe that there are ways in which this framework could be further strengthened. In particular, although it has made valuable contributions to the understanding of linguistic constraints on L2 variation (Level 2 of the psycholinguistic model of interlanguage variation), we focus this state-of-the-scholarship article on an issue that would benefit from more rigorous attention than it has received in previous research: social variation (i.e., Level 1 of the model). Variationist SLA research has given considerably more attention to linguistic constraints or Level 2 variation (Geeslin, 2011, pp. 501–502). As we discuss in Section 3 and Section 4, this focus is limiting because it does not reflect more recent developments in variationist sociolinguistics and other socially oriented approaches to SLA. However, we aim to initiate a call for a more concentrated focus on social (and extralinguistic) variation because in order to develop a stronger theoretical understanding of the ways in which social variation shapes learner development, renewed attention to the extralinguistic nature of language learning and variation is needed.

2.2. Previous L2 Variationist Research on Social Factors

In traditional variationist SLA, Level 1 of the psycholinguistic model of interlanguage variation, although termed social by Preston (2000), has been interpreted broadly to encompass extralinguistic variation. Researchers have investigated social factors that are commonly examined in sociolinguistics like gender and social class, as well as other individual and contextual factors like learning context, L2 proficiency, and topic. Focusing specifically on work that adopted the variationist approach, we review each of these variables in turn.

We begin with the gender of the learner. While not all studies that have investigated this factor have found that it conditions variation (e.g., present progressive in Spanish, Geeslin & Fafulas, 2012), various L2 investigations have shown that men use informal variants more than women, findings that were similar to native-speaker patterns. For example, Adamson and Regan (1991) investigated variable (ing) in words like “going”, “tempting”, and “nothing” (p. 4) among additional-language learners of English (native speakers of Cambodian or Vietnamese). Result showed that the male learners favored the informal variant [in] and women favored the formal variant [iŋ]. Mougeon et al. (2010), moreover, examined immersion learners in Canada and variation between two lexical items that mean “only” in French. They observed that men favored the informal variant juste, while women favored the formal seulement. In another study on French, Regan et al. (2009) investigated Irish learners who were studying abroad in France. They found that men favored the informal first-person-plural subject pronoun on “we” and women favored the formal variant nous. In contrast, however, Regan et al. also found that women deleted/l/in subject clitic pronouns more. In this case, women favored the informal variant more than men, which aligned with trends among native speakers.

Social class is another extralinguistic factor that has received attention in variationist SLA, though it has been investigated more often in L2 French than other languages. Regan et al. (2009), in their study of additional-language learners of French studying in France, found that middle-class participants were more likely to use the informal first-person-plural subject pronoun on, whereas upper class learners favored the formal nous. In Mougeon et al.’s (2010) work on immersion learners of French, social class did not constrain some of the variable structures under investigation (e.g., future-time reference, lexical items for car). However, when class did influence use, middle-class participants favored formal variants (e.g., retention of the preverbal negative particle ne and the pronoun nous) and upper-working class learners favored the more informal variants (omission of ne, pronoun on). Collectively, these two investigations on L2 French suggest that learners who belong to higher social classes use formal variants more often than those who belong to lower social classes.

Other factors that have received attention in variationist SLA include, for instance, learning context, L2 proficiency, and discourse topic. Regarding learning context, researchers examine whether the location where learners are exposed to the target language is connected to variable language behavior. One such study is Kanwit and Solon (2013), who investigated future-time reference in Spanish among two groups of learners: one who studied in Spain and another in Mexico. A group of native speakers living in the same cities where the learners were studying also participated in the study. The two native-speaker groups exhibited certain differences in their selection of future-time verb forms on a written contextualized preference task. For instance, whereas temporal distance (i.e., how far in the future an action is to occur) was a linguistic factor that influenced selection of future-time verb forms for the Spaniards, this factor was not significant for the Mexicans. Kanwit and Solon also found that over the course of the seven-week study abroad program, the learners in each location changed their variable patterns of verb selection in contexts of future-time reference, moving closer toward the patterns exhibited by the native speakers in the same region (i.e., comparatively greater preference for the morphological future in Spain). Whereas Kanwit and Solon (2013) constitutes an example of between-learning-context variation, research that investigates within-learning-context variation also exists. For instance, Wirtz and Pfenninger (2024) utilized virtual reality data-collection techniques to study learners’ use of dialectal varieties in German (i.e., standard German, Austrian dialect, and mixed varieties).8 They assigned each participant a score for their self-reported standard German exposure and their self-reported Austrian dialect exposure. Their analysis of these variables showed, for example, that only those participants who reported higher than average exposure to the Austrian dialect used this regional variety. Collectively, these two investigations indicate that learners gain sensitivity to the linguistic features that are present in a given learning context, and they can incorporate them into their language behavior.

Regarding L2 proficiency, the analysis of this variable has been particularly common in cross-sectional research on L2 Spanish, where the course level of instructed learners has been used often as a general means of distinguishing proficiency levels. For example, in Gudmestad’s (2012) investigation of mood distinction (i.e., the subjunctive–indicative contrast) and Kanwit’s (2017) study of future-time reference, the researchers identified stages of development for the variable structures under investigation by documenting differences in patterns of use across course levels. More recently, other assessments of proficiency have been employed in variationist SLA. In Solon and Kanwit (2022), for instance, participants completed an elicited imitation task, an instrument that has been used in various subfields of SLA as a general measure of proficiency (see Bowden, 2016 for information on the task; and Gudmestad & Edmonds, 2023, for the use of this proficiency measure in the study of variable subject expression). The researchers conducted a hierarchical cluster analysis on the scores, which identified three distinct proficiency levels. In the authors’ investigation of intervocalic/d/in Spanish (in words like cada “each”) in an aural contextualized preference task, Solon and Kanwit found that L2 proficiency on its own did not predict whether learners selected the deleted or realized variant. However, L2 proficiency significantly interacted with other independent variables. For example, the interaction between L2 proficiency and the preceding vowel (i.e., the vowel that came before/d/in the word containing intervocalic/d/) demonstrated that “when the preceding vowel was/i/rather than/a/, the low and mid groups selected deletion significantly less than the high group” (Solon & Kanwit, 2022, p. 815).

Other investigations have examined the topic of the interview (Regan et al., 2009) or topic seriousness as a means to examine sociostylistic variation, where certain topics are linked with (in)formality in language. Regarding topic seriousness, Donaldson (2017) studied verbal negation in French, specifically retention versus omission of the preverbal particle ne. He found that near-native speakers of French were more likely to use ne, the formal variant, when the topic was serious. In another analysis of the same speakers, Gudmestad et al. (2020) found that in contexts of future-time reference, the near-native speakers were also more likely with serious topics to use the inflectional future, again the formal variant, compared to the periphrastic future and present indicative.

Having reviewed some of the extralinguistic variables that have been investigated in this area of scholarship, we now turn to future directions for the study of the social side of the acquisition of sociolinguistic variation. We begin by reviewing the evolution of variationist sociolinguistics, which has been described as occurring in three waves, and highlight the need for L2 researchers to pursue more research that aligns with the second and third waves. We then consider L2 work conducted in frameworks other than variationism that have shed important light on the role of extralinguistic variation in L2 development. Such scholarship and L2 variationist research are not necessarily in active dialogue with each other, and we argue that greater consideration of the social dimension of language from L2 studies outside of variationism can help reveal understudied aspects of L2 variation.

3. Three Waves of Variationist Sociolinguistics

As previously mentioned, variationist SLA has drawn conceptually and methodologically from variationist sociolinguistics. This research strand began with William Labov’s (1972, inter alia) work on phonological variation in US English and led to a rich line of inquiry. Penelope Eckert (2012) has described variationist sociolinguistics as evolving over the course of three waves in the way it investigates language change and social meaning. Methodologically, first-wave studies have aimed to gather data from representative samples of a community, and quantitative analyses then identify links between linguistic variation and pre-established, macro-level demographic factors like age, gender, and socioeconomic class (Eckert, 2012, p. 90). The L2 studies described in the previous section align most closely with the first wave.9 Despite the valuable knowledge that has emerged from first-wave investigations, Eckert (2012, p. 90) observed that these studies “interpreted the social significance of variation on the basis of a general understanding of the [social] categories that served to select and classify speakers rather than through direct knowledge of the speakers themselves and their communities”. The second wave, in response, represents a departure from a priori social classifications and instead uses ethnographic methods to understand the local significance (i.e., tangible issues, locations, people) of social categories in a community and how local features shape language variation and change, with language users attributed as agents making decisions to use particular forms according to local practices and ideologies, rather than simply as a result of their macro-demographic characteristics (Eckert, 2012, pp. 90–91). For more on the local, see the notion of communities of practice in seminal works by Eckert and McConnell-Ginet (1992, 2007).

A well-known example of second-wave variationist sociolinguistics is Eckert’s (1989a, 2000) study of sound change among adolescents in a suburban US high school. Eckert spent two years at pseudonymous Belten High School conducting participant observation of and interviews with students. She focused on the final two years of high school of one graduating class. The research identified Jocks and Burnouts, two categories of students that had local significance. The contrasts between these two local social categories are manifested in various ways (see

Table 1. Differences between local social categories (adapted from Eckert, 1989a).

Characteristic	Jocks	Burnouts
Clothing	Pastel-colored	Dark-colored
Smoking habits	Do not smoke cigarettes	Smoke cigarettes
School activities	Involvement in school sports and clubs	Rejection of school extracurriculars
Spatial orientation	Occupy school facilities (eating lunch in the cafeteria, storing belongings in lockers)	Prefer school spaces less central to school life (courtyard, parking lot)

). Eckert (1989a, pp. 67–68) also observed clear differences in the two groups’ use of language, with one example being that Burnouts use double negation (e.g., You can’t do nothing) far more often than Jocks. While a detailed discussion of Eckert’s analysis of sound change is beyond this article’s scope, we highlight two relevant findings that emerged because of the focus on local social categories. For one, Eckert (1989b) observed differences in sound change between gender identities and between category memberships (i.e., Jocks and Burnouts) and, importantly, found that these two social variables interact such that “girls are asserting their category identities through language more than are the boys” (p. 265). Another observation is that the a priori classification of the adolescents according to their parents’ socioeconomic status did not account for their involvement in sound change, but their membership in the locally relevant categories of Jocks and Burnouts did (Eckert, 2018, p. 58). Taken together, these findings attest to the value of ethnographic work that seeks to identify social categories that are meaningful to the community under investigation and that subsequently applies these categories to the study of language variation and change.

To date, L2 variationist investigations generally do not classify themselves as examples of second-wave research, which suggests that there are many ways in which it could be applied to the study of L2 sociolinguistic competence.10 This work would begin with the identification of communities in which learners participate and the determination of relevant social categories based on deep local knowledge gained from interacting with and observing groups of and individual learners. For example, Bayley and Tarone (2012) note that the L2 classroom may form a community. Ethnographic studies on L2 classrooms could stand to make unique contributions to second-wave research because “L2 learner identity and group orientation may be defined in terms of categories not traditionally used in sociolinguistics; for example, L2 learners entering academic disciplines may define themselves more in terms of membership in academic discourse communities than membership in traditional speech communities” (p. 44). Investigations of L2 users living in the target-language context (e.g., study abroad participants, immigrants) would also be valuable because they may form local categories that have social meaning.11 Study abroad participants, for instance, may create and understand social meaning within their temporary residence in the community. Once researchers have identified locally relevant categories, they can be applied to the study of the L2 development of variable structures.

While the first wave accounted for macro-demographic factors and the second wave highlighted language users’ agency and locally meaningful social categories, the third wave is marked by its concern with speakers’ stylistic perspective and fluid expression of social meanings (see Eckert, 2018; Hall-Lew et al., 2021). Namely, as language users, we place ourselves in the social landscape through particular stylistic practices (i.e., ways of interacting) that are achieved in real time (Eckert, 2012, pp. 93–94). As members of particular social groups use certain forms, variable language usage becomes reinterpreted and fluid social meanings (e.g., associations with men, nerds, etc.) continue to become recombined in a bricolage of sorts (Hebdige, 1979; Lévi-Strauss, 1966; Kiesling, 2024, Chapter 6). In other words, through language, among other semiotic resources (e.g., clothing, gestures), individuals index ways of belonging or attempt to take on what they view as desirable traits. These practices and features (i.e., the bricolage) associated with particular linguistic or other semiotic usage become ideologically linked—together constituting an indexical field, or grouping of ideologically related meanings (Eckert, 2008; Campbell-Kibler, 2011)—and any of the social meanings can be invoked at a particular time based on ideological connections. Thus, the social meanings of linguistic forms are not fixed or static and instead include a range of possible meanings (i.e., an indexical field, as exemplified in the paragraph that follows, taking place within an ever-changing ideological field; Silverstein, 2003).

Social differentiation develops over time (e.g., through schooling, work) and individuals continue to develop distinctive styles. As language users index particular social meanings, individual and group styles develop and become fortified, with speakers taking repeated stances in their relationship to each other in an interaction and alignment (or not) with their interlocutor (see Kiesling, 2024). Registers are thus not only stylistic resources at a speaker’s disposal but also the end products of the bricolage of the multiple social meanings conveyed when speaking a certain way. Linguistic forms may then become associated with a particular social context (e.g., location) or type of speaker, and ideology becomes embedded in language and is constructed and reconstructed, revealing an ideological link between the linguistic and the social (Eckert, 2008; Irvine & Gal, 2000).

Consequently, in the third wave, style becomes viewed as evidence of choice or agency rather than simply the result of an individual’s membership in a particular sociodemographic group. Ideologies become revealed, as stylistic moves stem from how individuals view and interpret the surrounding world, how they hold and re-form ideologies, and how they position themselves in the social world. Thus, stylistic moves connect language to demographic categories in local practice, as individuals interpret and produce a range of styles (Eckert, 2008).

For instance, in his third-wave case study, Podesva (2007) examined the speech of “Heath” across different communicative contexts via acoustic measures and analyses of discourse functions of phenomena such as hyperarticulation, creaky voice, and falsetto. Hyperarticulation has been documented as indexing a range of social types, styles, or voices (e.g., aspirated intervocalic /t/ can invoke identity as a nerd(y) girl, a gay man, an Orthodox Jew, a speaker of British English, a school teacher), and thus in real time a gay man (i.e., Heath) has the agency to use longer bursts of /t/ aspiration to convey the persona of a “playful diva” at a social barbecue with friends (p. 4). In the first place, hyperarticulation becomes ideologically linked to these social types because it is associated with certain qualities (e.g., articulateness, elegance, educatedness, prissiness) that might be expected for a particular type of person. Moreover, in a given moment, a speaker may decide to hyperarticulate in order to take a particular stance (e.g., to be formal, polite, clear, emphatic, exasperated, angry, annoyed). It is precisely, then, the indexical connection between hyperarticulation and a particular form of gayness that makes Heath’s intentions clear to his audience of friends and enables him to play with (i.e., vary) language accordingly, making these features noticeable. These connections can be visualized in Figure 1 (see also Figure 4 of Eckert, 2008, p. 469). The particular qualities may be ideologically associated with multiple social types, and stances may be linked to a variety of different qualities. For instance, a school teacher and a nerdy girl may be associated with articulateness, and someone articulate and someone elegant may be linked to politeness. Thus, a range of social types, qualities, and stances come to form the indexical field of /t/ release. Some qualities may be more closely related and are therefore visualized more closely; the same is true for some stances.

Furthermore, language is but one semiotic resource available to individuals, so much like how linguistic forms may become associated with particular speakers, groups, and styles, so too can other meaningful resources (e.g., makeup, hair styles). Therefore, through the second and third waves, we see that language users are not passive members of certain social categories constrained by a particular variable but are agents expressing a certain meaning in a particular moment. The specific moments of language use by a particular speaker may generalize to a bigger picture, with social meanings interpreted and reinterpreted, becoming associated with certain groups and styles. Speakers then meaningfully play with such knowledge for their own communicative purposes.

Although the third wave shares with the second wave that it has been much less invoked in L2 variationist research than the first wave, there is a small but growing body of research that investigates how L2 speakers perform stylistic practices and reveal ideologies in real time (for recent overviews, see Smakman, 2022; Starr, 2023). For instance, Grammon (2024a) analyzed “Rita”, a learner studying abroad in Cuzco, Peru, and identified two principal phases through which she developed sociolinguistic competence. First, Rita gained interpretive abilities regarding regional forms and their social indexicality (i.e., connection to particular social types) in making first-order (i.e., direct) connections between forms and social meanings. She made ideological connections based on opinions regarding comprehensibility, prestige, and an idealized standard language in generally associating Cuzco Spanish as less complex and correct than the classroom Spanish to which she had been previously exposed. For instance, her host mother’s usage of hay que “one must” instead of tener que drew her attention as a local form generally absent from her prior instruction. In the second phase, Rita justified these connections through additional social experiences and began to connect regional forms with higher-order social-indexical values via representations of local social types (of Peruvians). We give two variable structures as examples. Rita came to associate usage of the familiar second-person singular tú (rather than more formal usted) with bricheros (e.g., catcallers and inappropriate cab drivers) and usage of the double possessive construction (e.g., the phrase-initial possessive pronoun in su mamá de Juan “the mother of Juan” rather than the standard definite article in la mamá de Juan), which she came through a raciolinguistic ideology to associate with cholos, or individuals originating from indigenous communities and typically able to speak Quechua. Thus, third-wave research has the ability to demonstrate how learners move from connecting forms (1) directly to social meanings (e.g., that a certain variety is prestigious) and then (2) to higher-order indexical types of local social relevance (e.g., that possessive pronoun usage might be associated with ties to an indigenous community in a particular regional setting).

4. Other Socially Oriented Research in SLA

L2 variationism has considered social factors less robustly than certain approaches outlined in the current section. Critically, the work described here has much to offer L2 variationism because it provides an interpretive lens for the analysis of social meaning. L2 research that does not classify itself as variationist in orientation has considered a range of social factors that have heretofore been underexplored in variationist SLA, including variables related to gender and sexuality; culture, race, and ethnicity; and interlocutor factors; among others. Critically, research described in the current section sheds light on how language users demonstrate agency in making decisions to use or avoid particular variants and varieties for social purposes. For each set of social variables in this section, we consider illustrative examples of recent and/or seminal research from non-variationist perspectives and highlight the ways in which extending this work to variationist SLA could be valuable for advancing knowledge about sociolinguistic competence.

Recent research on L2 development has afforded an important role to learner gender and sexuality that has tended to be absent from L2 variationist studies; we offer two examples. First, interdisciplinary work by Knisely (2020) reports attitudes toward non-binary forms in L2 French, drawing on frameworks of sociocultural theory (Lantolf et al., 2020), sociocultural gender diversity, trans studies, sociology of language, power, and identity, and raciolinguistics (see Alim et al., 2016). Knisely reveals the complex task of non-binary individuals in attempting to use forms perceived to be true to their own identities (i.e., not necessarily represented by pre-existing grammatical forms or the combination thereof) while still being comprehensible to interlocutors from a range of experience levels with non-binary individuals. For instance, the hybrid pronoun iel (built off il “he” and elle “she”, equivalent to English singular “they”) was the highest rated of the innovative pronouns, and inclusive punctuated suffixes (e.g., use of combined affixes as in gentil.le for “kind”, integrating masculine gentil and feminine gentille) also received high comprehensibility scores. L2 variationist work could benefit from approaches such as Knisely’s, which move beyond the binary in analyses of learner gender identity and its relationship with sociolinguistic variation.

Second, recent interdisciplinary research by Moore (2019) has provided vivid insights into the linguistic experiences of LGBT learners of L2 Japanese, drawing on three theoretical frameworks: theory of social domains (i.e., that the social world is made up of psychobiography, situated activity, social setting, and contextual resources), sexual identity management (i.e., that stigmatized identities must be actively managed, e.g., Nelson, 2010), and the model of social participation trajectories (i.e., that inbound (new) trajectories may differ from insider (ongoing, established) ones). Moore’s qualitative study demonstrates how learners come to indicate nonheteronormative identities in class through salient indicators, insider evidence, and explicit statements, illustrated in

Table 2. Cues affecting learners’ decisions to indicate nonheteronormative identities (adapted from Moore, 2019).

Cue	Description	Example
Salient indicators	Possible signifiers of classmates’/teachers’ acceptance of LGBT individuals	Status as young, a woman
Insider evidence	Comments and actions regarding acceptance of LGBT individuals	Facial expressions, use of outdated terms
Explicit statements	Overt declaration of acceptance	Pre-semester survey with such a statement

. Moore’s work, like that of Knisely, benefits from being informed by multiple social theories. As we explore in the next paragraph, such interdisciplinary grounding could help subsequent L2 variationist work to predict and interpret language variation realized via learner interaction.

Subsequent work on sociolinguistic competence from a variationist perspective might, for example, use the powerful social frameworks and qualitative elucidations from Moore (2019) to consider the possible role of nonheteronormative gender identity in constraining other forms of language variation in the presence of independent linguistic and/or social variables and how such constraints may change over time as the learner becomes more comfortable with the instructor, classmates, and other interlocutors according to the strategies outlined above. For instance, in a study focused on second-person singular address forms in German (i.e., traditionally informal du and traditionally formal Sie), in addition to noting the relevant cues identified by Moore, the researcher could analyze how a learner’s gender identity shapes how they address their significant other and their instructor. Adding to the complex considerations of such a project, the German address system has been argued to be more of a continuum than a binary distinction of formality (Hickey, 2003), which could be conceptualized as a scale of perceived hierarchical and social distance (Kretzenbacher, 2010). The researcher could also code other contextual forms known to affect the acquisition and use of forms of address (e.g., context of interaction, level of imposition in any requests, gender identities of interlocutors). The analyst could then see to what extent the address form used for the partner matches the learner’s acquisition of forms of address more generally or whether the learner meaningfully manipulates person reference for certain types of interlocutors or micro-contextual factors. This information could be complemented qualitatively by the three cues noted by Moore. For example, with interlocutors who differ in terms of salient indicators, to what extent does the learner utilize different forms of address? Does insider evidence that reveals an interlocutor to be more/less accepting correlate with diverging rates of pronoun usage? Moreover, conceptualization of address systems as scalar/continuous rather than binary would be consistent with the cognitive–functional origins of variationism in not forcing dichotomous binaries of categorization (Tomasello, 1998), with more nuanced conceptualizations of other dimensions of social life (e.g., non-binary considerations of gender, Sauntson, 2020), and with statistical suggestions to not impose categorical distinctions on continuous/scalar constructs (Baayen, 2014).

Moore (2019) and Knisely (2020) point to trends elucidated across a range of methodologies, which have at times revealed differential acquisitional patterns based on the role of identity work or access to differential types of input (e.g., Cashman, 2021; Nelson, 2010). Namely, because a learner may perform their gender identity differentially in a given moment (based on the interlocutor, communicative goal, etc.) and because a learner’s access to input may differ based on their gender identity and the attitudes of potential interlocutors, variationist SLA would do well to prioritize such information in future studies. As seen in the prior example, a learner’s fluctuation in rates of familiar or formal pronoun usage may, in fact, reveal a conscious decision to express themselves in a certain way depending on perceived comfort or hesitation with the audience rather than, for instance, inconsistency in an ability to maintain consistency in forms of pronominal address in German. Similarly, more in-depth information about a learner’s identity and orientation toward Peruvian culture, à la Grammon (2024a), could help shed light on their variability in the use of second-person pronouns and double possessives. Such information will help inform the researcher whether a learner is continuing to acquire variable structures in determining when more (in)formal pronouns or possessive constructions might be expected but also when a learner decides to (not) use a given form due to a desire to identify with a given interlocutor. This future work could complement Grammon’s qualitative examples with larger-scale production data from learners to see to what extent more widespread usage supports or diverges from the forms we might expect a learner to use given a particular social and linguistic context and a particular orientation to the target culture and interlocutor.

L2 research outside of variationism has also shed important light on the roles of culture, race, and ethnicity on L2 experience and development. Beginning with culture, Lybeck (2002) used acculturation theory (i.e., that social factors play an essential role in influencing the quantity and quality of input) to demonstrate links between learners’ cultural identification and development of L2 pronunciation for English-speaking American learners of L2 Norwegian studying in Norway. Namely, learners who showed greater identification with Norwegian culture developed more target-like development of pronunciation of/r/. Thus, as we have seen with second-wave sociolinguistics, Lybeck used knowledge of learners’ context of social interaction (i.e., their attitudes toward Norwegian culture) to help explain language variation. Perception of culture has also received attention, as in perception of learner approximation to the target culture in Gatbonton et al. (2005), which found that Chinese- and French-speaking learners of English in Quebec perceived that L2 speakers who sounded more like target-language speakers of English were less loyal to their home ethnic group. Work such as that by Gatbonton and colleagues helps reinforce the point that learners are agents who may decide to approximate local target-language variants because they want to sound like (native) speakers from the region or they may consciously adjust their speech to maintain affiliation with another community (e.g., that of their L1). Such studies add important explanatory power for the role of social variables in constraining L2 use and perception because the theories from which they draw help to provide an interpretative lens for data and help make predictions regarding the roles of identity factors and other contextual and situational elements.

Moreover, the roles of race and ethnicity in constraining exposure, usage, and development in the L2 tend to be underexplored, based on emphasis on college-age students who are overrepresented by Caucasian/European individuals who may not strongly identify with a particular cultural background. Nevertheless, Gatbonton et al. (2005), Grammon (2024a), Knisely (2020), and Moore (2019), along with recent efforts in raciolinguistics (Alim et al., 2016), push SLA forward by considering underrepresented perspectives. For example, through the lens of transformative socialization (i.e., how learners and their communities enact new communicative selves in collaboratively co-constructing and negotiating multiple identities, Anya, 2017, p. 4), Anya documented in detail the experiences and linguistic development of African–American students studying abroad in a majority Black (i.e., Afro-Brazilian) region of Brazil, noting for instance positive learner reports of sojourning in an environment of greater understanding and comfort with respect to racial identity. Subsequent research from a variationist perspective could analyze a dataset like the one in Anya (2017) for instances of variable structures that may develop longitudinally during the sojourn, as learners studying abroad can be exposed to the rich input and socialization detailed by Anya. In particular, future work could investigate whether learners who report greater comfort in the host country more closely approximate local Brazilian features such as palatalization of /t/ that typically occurs in regions of Brazil.

Finally, also important along the social dimension are interlocutor factors, such as social distance, power differential, and (un)familiarity. Non-variationist research from the field of L2 pragmatics has contributed knowledge on the multiplicity of these factors. For example, learners make more formal requests when interacting with interlocutors in contexts of greater social distance, unfamiliarity, cultural differences, and imposition, and when the interlocutor is in a position of greater power than the learner (for a comprehensive overview, see Félix-Brasdefer & Shively, 2021). One example is Kobayashi and Rinnert’s (2003) findings for English learners in Japan, who made more supporting moves (e.g., preparators like “May I ask you a favor?”) and delayed requests (i.e., placing a request later in the discourse) when the level of imposition on the interlocutor was higher. Nevertheless, L2 sensitivity to these factors may vary, as the same study revealed that other learners may not necessarily use more formal language in the face of these factors. For instance, high-proficiency learners showed greater use of requests with informal “want” when the level of imposition was higher. It is worthwhile to consider the range of interlocutor factors (e.g., level of imposition, relationship with interlocutor) when investigating social effects on variable structures, as the identities of each interlocutor play an important role in shaping the communicative context, how communication unfolds, and the identity work performed by each participant in the discourse. In this vein, the study of variable structures could draw on L2 pragmatics research in order to incorporate variables that pertain to the relationship between learners and their interlocutors. This would dovetail with Geeslin’s (2020) variationist model of learner/interlocutor interaction, which places the relationship between the learner and the interlocutor at the center of language development and use of language variation. This model builds on Preston’s (2000) psycholinguistic model of interlanguage variation by showing how individual characteristics can shape how interlocutors are perceived and how speakers use language in specific moments of interaction.

In sum, this section offers a snapshot of some of the ways in which non-variationist SLA has advanced the understanding of how social factors shape acquisition. We believe that extending this body of work’s interpretive lenses and insights to variationist SLA’s exploration of sociolinguistic competence has the potential to contribute significantly to knowledge about the development of variable structures.

5. Conclusions

We began the current paper by offering an overview of variationist SLA and identifying some of the extralinguistic factors shown to be important for the L2 development of variable structures. Despite the valuable contributions of previous work, we highlighted that variationist SLA has paid greater attention to the role that linguistic factors play in the acquisition of variable structures and that there is thus much we still have to learn about how the extralinguistic side of language shapes this aspect of sociolinguistic competence (Geeslin, 2011, pp. 501–502). We then reflected on the evolution of variationist sociolinguistics and looked to other socially oriented approaches to SLA in order to establish an agenda for future variationist SLA scholarship. Throughout this discussion, we identified specific ways in which variationist SLA researchers could draw on insights from these bodies of research in order to incorporate greater emphasis on social variation into the study of L2 sociolinguistic competence. We believe these recommendations fall into three general issues for an agenda for future studies, listed in (i) through (iii):

(i)

The second and third wave of sociolinguistics and contributions from non-variationist L2 approaches show us that a critical aspect of language usage and development pertains to language users and learners themselves—the active role they play in their language behavior and the agency they have to use language to reflect their identity. For instance, with approaches informed by sociocultural theory, van Compernolle and Williams (2012) have shown how learners use variation to perform and negotiate their identities in an L2 and how the meanings they assign to variation may change over time for variables such as first-person plural and second-person singular reference and variable use of negative ne in L2 French.12 Variationist SLA has much to learn about L2 sociolinguistic competence by better incorporating the constructs of agency and identity into the investigation of variable structures and by examining these constructs along with other (extra)linguistic factors. This work could also help reveal how learner language ideologies develop and the extent to which these reflect or diverge from ideologies present in the target-language culture. In turn, how learners come to associate variable linguistic forms with locally relevant social types and how additional orders of indexicality unfold can continue to inform what we know about L2 development.

(ii)

Most L2 variationist research has employed quasi-experimental techniques and has been quantitative (see Regan, 2023; Wirtz et al., 2024, for recent examples of qualitative studies). While analytical tools like regression models remain valuable, the evolution of variationist sociolinguistics and other socially oriented approaches to SLA has demonstrated that ethnography and qualitative analyses shed important light on the intricacies of social factors (e.g., local categories like Burnouts and complex variables like gender identity). Thus, diversifying the research methods employed in variationist SLA appears to be a necessary step for expanding the study of the relationship between the extralinguistic nature of language and the L2 development of variable structures (see Riazi & Farsani, 2024, for a discussion of mixed-methods research in applied linguistics more generally). The aforementioned recent example of the use of virtual reality in Wirtz and Pfenninger (2024) illustrates another way in which new methods may be used to untap the role of social characteristics in language development, and the study’s dense data collection points help to reveal how sociolinguistic competence develops over time. Such repeated elicitations will contribute social information to a base of studies in L2 variationism that have tended to follow the general applied linguistics trend of largely including one-time cross-sectional or relatively limited longitudinal sampling (Ortega & Byrnes, 2008).

(iii)

Variationist SLA research has primarily investigated Caucasian college students. These participant pools come from Western, educated, industrialized, rich, and democratic (i.e., WEIRD) societies, which are not representative of the extent of human diversity, thus raising questions about the generalizability of the knowledge that has emerged from research on this population (Henrich et al., 2010). Broadening the study of variable structures to other learner populations (see Anya, 2017) will therefore enable researchers to more fully investigate the role that social factors play in the development of sociolinguistic competence because, for one, the social and linguistic diversity that exist in the world will be more accurately represented.

Our hope is that this state of the science of and agenda for future scholarship on variationist SLA offers variationist researchers useful insights for more fully integrating the social dimension of language into their investigations of L2 sociolinguistic competence.